@getanyapi/sdk 0.30.0 → 0.31.0

package/dist/index.d.ts

@@ -488,6 +488,11 @@ interface AhrefsBacklinksInput {
      * Default: subdomains.
      */
     mode?: "exact" | "subdomains";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The domain or page URL to find backlinks for (e.g. "ahrefs.com").
      */
@@ -546,6 +551,11 @@ interface AhrefsKeywordIdeasInput {
      * The seed keyword to expand into related suggestions (e.g. "coffee").
      */
     keyword: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface AhrefsKeywordIdeasItem {
     /**
@@ -608,6 +618,11 @@ interface AhrefsKeywordsInput {
      * The search term to analyze (e.g. "seo tools").
      */
     keyword: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface AhrefsKeywordsItem {
     /**
@@ -647,6 +662,11 @@ interface AhrefsOverviewInput {
      * Default: subdomains.
      */
     mode?: "exact" | "subdomains";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The domain or page URL to analyze (e.g. ahrefs.com).
      */
@@ -809,6 +829,11 @@ interface AirbnbSearchInput {
      * Range: minimum 0.
      */
     pets?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Maximum search price in the selected currency (e.g. 300).
      * Range: minimum 0.
@@ -899,6 +924,11 @@ interface AlibabaSearchInput {
      * Range: minimum 1, maximum 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keywords to search for on Alibaba (e.g. "bluetooth speaker wholesale").
      */
@@ -998,6 +1028,11 @@ interface AmazonAsinsInput {
      * Range: minimum 1, maximum 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface AmazonAsinsItem {
     /**
@@ -1056,6 +1091,11 @@ interface AmazonBestsellersInput {
      * Default: 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Amazon Best Sellers category URL (e.g. https://www.amazon.com/Best-Sellers-Electronics/zgbs/electronics).
      */
@@ -1121,6 +1161,11 @@ interface AmazonBestsellersData {
  * Input for Amazon Product (amazon.product).
  */
 interface AmazonProductInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Amazon product URL (e.g. https://www.amazon.com/dp/B0CX23V2ZK).
      */
@@ -1222,6 +1267,11 @@ interface AmazonReviewsInput {
      * Default: 50.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Amazon product ASIN or full product URL (e.g. B07CMS5Q6P).
      */
@@ -1304,6 +1354,11 @@ interface AmazonSearchInput {
      * Default: 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Amazon search or category URL to pull results from (e.g. https://www.amazon.com/s?k=gaming+mouse).
      */
@@ -1437,6 +1492,11 @@ interface ApolloOrganizationInput {
      * Organization identifier returned by an Apollo organization endpoint.
      */
     organizationId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Apollo Organization (apollo.organization).
@@ -1572,6 +1632,11 @@ interface ApolloOrganizationEnrichInput {
      * Organization domain without a path, such as apollo.io.
      */
     domain: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Apollo Organization Enrichment (apollo.organization_enrich).
@@ -1707,6 +1772,11 @@ interface ApolloOrganizationJobsInput {
      * Organization identifier returned by an Apollo organization endpoint.
      */
     organizationId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface ApolloOrganizationJobsJob {
     /**
@@ -1797,6 +1867,11 @@ interface ApolloOrganizationNewsInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface ApolloOrganizationNewsArticle {
     /**
@@ -1871,6 +1946,11 @@ interface ApolloOrganizationsBulkEnrichInput {
      * Organization domains to enrich, with at most 10 domains per request.
      */
     domains: string[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The enriched organization for the domain at this index, or null when that domain had no match.
@@ -2033,6 +2113,11 @@ interface ApolloOrganizationsSearchInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface ApolloOrganizationsSearchOrganization {
     /**
@@ -2156,6 +2241,11 @@ interface ApolloPeopleSearchInput {
      * Person locations to match.
      */
     personLocations?: string[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Seniority levels to match.
      */
@@ -2293,6 +2383,11 @@ interface ApolloPersonEnrichInput {
      * Organization name used with the person's name.
      */
     organizationName?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface ApolloPersonEnrichEmploymentHistory {
     /**
@@ -2605,6 +2700,11 @@ interface AppstoreReviewsInput {
      * Range: minimum 1, maximum 100.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface AppstoreReviewsItem {
     /**
@@ -2682,6 +2782,11 @@ declare class AppstoreNamespace {
  * Input for Bluesky Post (bluesky.post).
  */
 interface BlueskyPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Bluesky post URL, e.g. "https://bsky.app/profile/bsky.app/post/3l6oveex3ii2l".
      */
@@ -2725,6 +2830,11 @@ interface BlueskyProfileInput {
      * Bluesky handle, e.g. "bsky.app" or "jay.bsky.team".
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Bluesky Profile (bluesky.profile).
@@ -2755,6 +2865,11 @@ interface BlueskyUserPostsInput {
      * Bluesky handle, e.g. "bsky.app" or "jay.bsky.team".
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface BlueskyUserPostsPost {
     /**
@@ -2866,6 +2981,11 @@ interface BookingSearchInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Destination city to search for stays in (e.g. Paris).
      */
@@ -2967,6 +3087,11 @@ interface CapterraReviewsInput {
      * Default: 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Capterra product URL, for example https://www.capterra.com/p/135003/Slack/. Capterra identifies a product by both its numeric id and its slug, so the full URL is required.
      */
@@ -3185,6 +3310,11 @@ interface ChatgptSearchInput {
      * Default: US.
      */
     country?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Question or research prompt for ChatGPT to answer using web search.
      */
@@ -3294,6 +3424,11 @@ interface CoinmarketcapListingsInput {
      * Range: minimum 1, maximum 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface CoinmarketcapListingsItem {
     /**
@@ -3407,6 +3542,11 @@ interface CompanyResearchInput {
      * Company name, resolved to a domain before the run starts. Mutually exclusive with domain and email.
      */
     name?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Company Research (company.research).
@@ -3579,6 +3719,11 @@ interface CompanyEnrichmentCrustdataV3Input {
     companyName?: string;
     exactMatch?: boolean;
     fields?: unknown;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface CompanyEnrichmentCrustdataV3HeadcountTimeserie {
     /**
@@ -3972,6 +4117,11 @@ interface CompanyEnrichmentLushaInput {
      * Company domain, e.g. posthog.com. Send the bare domain, without a scheme or www.
      */
     domain?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface CompanyEnrichmentLushaNaicsCode {
     /**
@@ -4207,6 +4357,11 @@ interface CompanyEnrichmentPeopledatalabsInput {
      * Company name, for when you have no domain.
      */
     name?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * People Data Labs boolean expression over top-level fields that a match must satisfy, e.g. website and industry.
      */
@@ -4573,6 +4728,11 @@ interface CompanyEnrichmentProspeoInput {
      * Company domain, e.g. stripe.com. The most reliable identifier.
      */
     companyWebsite?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface CompanyEnrichmentProspeoEvent {
     /**
@@ -4954,6 +5114,11 @@ interface CompanySearchAiArkInput {
      * Default: 0.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Maximum companies to return on this page.
      * Range: minimum 1, maximum 100.
@@ -5358,6 +5523,11 @@ interface CompanySearchCrustdataV3Input {
      * Default: 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     sorts?: CompanySearchCrustdataV3Sort[];
 }
 interface CompanySearchCrustdataV3Companie {
@@ -5869,6 +6039,11 @@ interface CompanySearchFullenrichInput {
      * Default: 0.
      */
     offset?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Filter by company specialty. Each entry is an object taking a `value` string and an optional `exact_match` boolean, e.g. [{"value": "stripe.com", "exact_match": true}]. A bare string is rejected.
      */
@@ -6027,6 +6202,11 @@ interface CompanySearchPeopledatalabsInput {
      * Default: 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Elasticsearch-style query over the People Data Labs company dataset, e.g. {"bool": {"must": [{"term": {"website": "posthog.com"}}]}}. Send this or sql, never both.
      */
@@ -6469,6 +6649,11 @@ interface CompanySearchProspeoInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface CompanySearchProspeoCompanie {
     /**
@@ -6909,6 +7094,11 @@ interface CompanySearchQuickenrichInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Filter on revenue band.
      */
@@ -7255,6 +7445,11 @@ interface CompanySearchTheirstackInput {
      * Page number. Required when using page-based pagination.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Return companies that have all of these fields not null. For example, if you pass ['domain', 'linkedin_url'], it will return companies that have both domain AND linkedin_url set.
      */
@@ -7697,6 +7892,11 @@ interface CongressTradesInput {
      * Range: minimum 1, maximum 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Earliest transaction date to include, inclusive, in YYYY-MM-DD format (e.g. 2026-01-01).
      */
@@ -7884,6 +8084,11 @@ interface DexscreenerTokensInput {
      * Sort direction: desc or asc (e.g. desc).
      */
     order?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Field to sort tokens by (e.g. volume, txns, liquidity, marketCap, trendingScoreH24).
      */
@@ -7988,6 +8193,11 @@ declare class DexscreenerNamespace {
  * Input for Douyin Profile (douyin.profile).
  */
 interface DouyinProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Douyin sec_user_id for the public account.
      */
@@ -8066,6 +8276,11 @@ interface DouyinSearchVideosInput {
      * Default: 0.
      */
     duration?: "0" | "0-1" | "1-5" | "5-10000";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Publication window in days. Use the canonical JSON integer 0 for any time, 1 for one day, 7 for seven days, or 180 for 180 days; legacy numeric strings remain accepted.
      */
@@ -8179,6 +8394,11 @@ interface DouyinUserPostsInput {
      * Default: 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Douyin sec_user_id for the public account.
      */
@@ -8272,6 +8492,11 @@ interface DouyinUserPostsData {
  * Input for Douyin Video (douyin.video).
  */
 interface DouyinVideoInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Public Douyin video share URL.
      * Format: uri.
@@ -8357,6 +8582,11 @@ interface DouyinVideoCommentsInput {
      * Range: minimum 0.
      */
     cursor?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Douyin aweme_id for the video.
      */
@@ -8508,6 +8738,11 @@ declare class DouyinNamespace {
  * Input for eBay Product (ebay.product).
  */
 interface EbayProductInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full eBay listing URL (e.g. https://www.ebay.com/itm/133576802017). The marketplace is taken from the host, so an ebay.co.uk or ebay.de URL returns that site's listing and currency. Item ids come back on every row of ebay.search and ebay.sold_listings.
      */
@@ -8633,6 +8868,11 @@ interface EbayProductData {
  * Input for eBay Product Full (ebay.product_full).
  */
 interface EbayProductFullInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full eBay listing URL (e.g. https://www.ebay.com/itm/133576802017). Works on ended and sold listings as well as live ones, so it composes with ebay.sold_listings: pull the comps cheaply, then enrich the few you care about.
      */
@@ -8780,6 +9020,11 @@ interface EbaySearchInput {
      * Range: minimum 0.
      */
     minPrice?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keywords, e.g. "nintendo switch" or "vintage levis 501".
      */
@@ -8871,6 +9116,11 @@ interface EbaySoldListingsInput {
      * Range: minimum 0.
      */
     minPrice?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keyword for sold items (e.g. iphone 13 pro).
      */
@@ -9046,6 +9296,11 @@ interface EmailFindInput {
          */
         surname: string;
     };
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface EmailFindItem {
     domain?: string;
@@ -9079,6 +9334,11 @@ interface EmailVerifyInput {
      * The email address to verify (e.g. jane.doe@acme.com). Exactly one @, a dotted domain, no whitespace or angle brackets. Addresses on reserved, never-deliverable TLDs (.invalid, .test, .example, .localhost, .local, .internal, .blink) and HTML/JSON escape artifacts (a u003e prefix) are rejected locally with no charge.
      */
     email: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface EmailVerifyItem {
     /**
@@ -9170,6 +9430,11 @@ interface EmailFindingDropleadsInput {
      * Person's last name.
      */
     lastName: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Finding - DropLeads (email_finding.dropleads).
@@ -9226,6 +9491,11 @@ interface EmailFindingHunterCountInput {
      * Company domain without a scheme or path, e.g. stripe.com.
      */
     domain: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Count only personal mailboxes belonging to a named person, or only generic ones such as info@ and support@.
      * One of: personal, generic.
@@ -9384,6 +9654,11 @@ interface EmailFindingHunterDomainInput {
      * Range: minimum 0.
      */
     offset?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only return contacts that carry this field.
      * One of: full_name, position, phone_number.
@@ -9546,6 +9821,11 @@ interface EmailFindingIcypeasInput {
     domainOrCompany: string;
     firstname?: string;
     lastname?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface EmailFindingIcypeasEmail {
     /**
@@ -9671,6 +9951,11 @@ interface EmailFindingQuickenrichInput {
      * Format: uri.
      */
     linkedinUrl?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Finding - QuickEnrich (email_finding.quickenrich).
@@ -9782,6 +10067,11 @@ interface EmailFindingZerobounceInput {
      * Middle name, when the company's address format uses one.
      */
     middleName?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Finding - ZeroBounce (email_finding.zerobounce).
@@ -9821,6 +10111,11 @@ interface EmailFindingZerobounceDomainInput {
      * Company domain to inspect, e.g. hubspot.com.
      */
     domain: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface EmailFindingZerobounceDomainOtherFormat {
     /**
@@ -9962,6 +10257,11 @@ interface EmailVerificationAllegrowInput {
      * Format: email.
      */
     email: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Verification - Allegrow (email_verification.allegrow).
@@ -10028,6 +10328,11 @@ interface EmailVerificationBouncebanInput {
      * One of: regular, deepverify.
      */
     mode?: "regular" | "deepverify";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Verification - BounceBan (email_verification.bounceban).
@@ -10058,6 +10363,11 @@ interface EmailVerificationIcypeasInput {
      * Format: email.
      */
     email: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Verification - Icypeas (email_verification.icypeas).
@@ -10151,6 +10461,11 @@ interface EmailVerificationZerobounceInput {
      * IP address the address signed up from. ZeroBounce uses it to add the geographic fields to the answer.
      */
     ipAddress?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Verification - ZeroBounce (email_verification.zerobounce).
@@ -10247,6 +10562,11 @@ interface EmailVerificationZerobounceActivityInput {
      * Format: email.
      */
     email: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Email Activity - ZeroBounce (email_verification.zerobounce_activity).
@@ -10342,6 +10662,11 @@ interface FacebookAdDetailsInput {
      * Meta Ad Library ad ID (e.g. "702369045530963"). Provide either id or url.
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Meta Ad Library ad URL (e.g. "https://www.facebook.com/ads/library?id=1185617869915074"). Provide either id or url.
      */
@@ -10460,6 +10785,11 @@ interface FacebookAdDetailsFullInput {
      * Meta Ad Library ad ID - the numeric id in an Ad Library URL (e.g. "1519158199783790" from https://www.facebook.com/ads/library/?id=1519158199783790).
      */
     id: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface FacebookAdDetailsFullCreative {
     /**
@@ -10632,6 +10962,11 @@ interface FacebookAdTranscriptInput {
      * Meta Ad Library ad ID (e.g. "1020359190509080"). Provide either id or url.
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Meta Ad Library ad URL (e.g. "https://www.facebook.com/ads/library?id=1020359190509080"). Provide either id or url.
      */
@@ -10682,6 +11017,11 @@ interface FacebookAdsSearchInput {
      * One of: ALL, IMAGE, VIDEO, MEME, IMAGE_AND_MEME, NONE.
      */
     mediaType?: "ALL" | "IMAGE" | "VIDEO" | "MEME" | "IMAGE_AND_MEME" | "NONE";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search the Meta Ad Library for (e.g. "protein powder").
      */
@@ -10829,6 +11169,11 @@ interface FacebookCommentRepliesInput {
      * The feedback_id of the comment (not the comment id).
      */
     feedbackId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface FacebookCommentRepliesReplie {
     /**
@@ -10923,6 +11268,11 @@ interface FacebookCompanyAdsInput {
      * Company's Ad Library page ID. Provide either pageId or companyName.
      */
     pageId?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Sort order: impressions (highest first, the default) or recent (most recent).
      * One of: impressions, recent.
@@ -11031,6 +11381,11 @@ interface FacebookEventDetailsInput {
      * The event's numeric identifier.
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The event's Facebook URL.
      */
@@ -11088,6 +11443,11 @@ interface FacebookEventsInput {
      * Pagination cursor from a previous response to fetch the next page.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Timeframe filter for the returned events. Defaults to all time.
      * One of: today, this_week, next_week.
@@ -11146,6 +11506,11 @@ interface FacebookEventsSearchInput {
      * Pagination cursor from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The query to search events for.
      */
@@ -11211,6 +11576,11 @@ interface FacebookFollowersInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Facebook page or profile URL to list follows for (e.g. https://www.facebook.com/nasa).
      */
@@ -11252,6 +11622,11 @@ interface FacebookGroupPostsInput {
      * Pagination cursor from a previous response to fetch the next page.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Ordering for the returned posts (e.g. TOP_POSTS).
      * One of: TOP_POSTS, RECENT_ACTIVITY, CHRONOLOGICAL, CHRONOLOGICAL_LISTINGS.
@@ -11341,6 +11716,11 @@ interface FacebookMarketplaceInput {
      * Longitude of the search location (e.g. '-97.7475').
      */
     lng: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Maximum listing price in whole currency units, e.g. 500 for $500. Facebook may mix in a few suggested listings outside the range.
      * Range: minimum 0.
@@ -11412,6 +11792,11 @@ interface FacebookMarketplaceItemInput {
      * Facebook Marketplace item ID.
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Facebook Marketplace item URL.
      */
@@ -11469,6 +11854,11 @@ interface FacebookMarketplaceItemData {
  * Input for Facebook Marketplace Location Search (facebook.marketplace_location_search).
  */
 interface FacebookMarketplaceLocationSearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Location search query (e.g. a city name).
      */
@@ -11524,6 +11914,11 @@ interface FacebookPageContactInput {
      * Facebook Page URL or page ID to look up (e.g. https://www.facebook.com/nasa).
      */
     page: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface FacebookPageContactItem {
     /**
@@ -11582,6 +11977,11 @@ interface FacebookPhotosInput {
      * Pagination cursor from a previous response to fetch the next page.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * URL of the public Facebook page or profile to fetch photos from (e.g. https://www.facebook.com/Spurs).
      */
@@ -11634,6 +12034,11 @@ interface FacebookPhotosData {
  * Input for Facebook Post (facebook.post).
  */
 interface FacebookPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Facebook post URL.
      */
@@ -11669,6 +12074,11 @@ interface FacebookPostCommentsInput {
      * Facebook feedback id for the post (alternative to url).
      */
     feedbackId?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Facebook post URL.
      */
@@ -11718,6 +12128,11 @@ interface FacebookPostCommentsData {
  * Input for Facebook Post Transcript (facebook.post_transcript).
  */
 interface FacebookPostTranscriptInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Facebook post or video URL.
      */
@@ -11738,6 +12153,11 @@ interface FacebookProfileInput {
      * Facebook page handle/username.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Facebook page URL.
      */
@@ -11772,6 +12192,11 @@ interface FacebookProfileEventsInput {
      * Pagination cursor from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Facebook page URL.
      */
@@ -11830,6 +12255,11 @@ interface FacebookProfilePostsInput {
      * Facebook page id.
      */
     pageId?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Facebook page/profile URL.
      */
@@ -11908,6 +12338,11 @@ interface FacebookProfileReelsInput {
      * Pagination cursor from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Facebook page/profile URL.
      */
@@ -11953,6 +12388,11 @@ interface FacebookProfileReelsData {
  * Input for Facebook Company Search (facebook.search_companies).
  */
 interface FacebookSearchCompaniesInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search advertiser pages for (e.g. "nike").
      */
@@ -12018,6 +12458,11 @@ interface FacebookSearchPagesInput {
      * Optional free-text location to narrow the search: a city, province, or country (e.g. 'Berlin').
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search Facebook Pages for (e.g. 'coffee roasters').
      */
@@ -12088,6 +12533,11 @@ interface FacebookSearchPostsInput {
      * Optional location to narrow results; include both city and country for best matches (e.g. 'Paris, France').
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword or phrase to search Facebook posts for (e.g. 'product launch').
      */
@@ -12530,6 +12980,11 @@ interface FiverrSearchInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Fiverr search or category page URL to extract gigs from.
      */
@@ -12629,6 +13084,11 @@ interface G2ReviewsInput {
      * Default: 50.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * G2 product slug, for example hubspot-marketing-hub. A full G2 product URL is also accepted and reduced to its slug.
      */
@@ -12787,6 +13247,11 @@ declare class G2Namespace {
  * Input for Gemini Search (gemini.search).
  */
 interface GeminiSearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Question or research prompt for Gemini to answer using web search.
      */
@@ -12849,6 +13314,11 @@ declare class GeminiNamespace {
  * Input for GitHub Repository (github.repository).
  */
 interface GithubRepositoryInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * GitHub repository URL (e.g. https://github.com/facebook/react).
      */
@@ -12949,6 +13419,11 @@ interface GithubTrendingDevelopersInput {
      * Programming language to filter trending developers (e.g. javascript, python, go).
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Trending range: daily, weekly, or monthly (defaults to daily).
      */
@@ -12996,6 +13471,11 @@ interface GithubTrendingRepositoriesInput {
      * Filter by programming language (e.g. "go", "typescript"). Omit for all languages.
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Trending window.
      * One of: daily, weekly, monthly.
@@ -13037,6 +13517,11 @@ interface GithubUserInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of GitHub User (github.user).
@@ -13113,6 +13598,11 @@ interface GithubUserActivityInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Year of contribution activity to return (defaults to the current year).
      */
@@ -13156,6 +13646,11 @@ interface GithubUserContributionsInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Calendar year of the contribution graph. Defaults to the current year.
      */
@@ -13209,6 +13704,11 @@ interface GithubUserFollowersInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface GithubUserFollowersFollower {
     /**
@@ -13256,6 +13756,11 @@ interface GithubUserFollowingInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface GithubUserFollowingFollowing {
     /**
@@ -13303,6 +13808,11 @@ interface GithubUserPullRequestsInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only include pull requests created on or after this date (YYYY-MM-DD).
      */
@@ -13369,6 +13879,11 @@ interface GithubUserRepositoriesInput {
      * GitHub username.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Repository ordering.
      * One of: created, updated, pushed, full_name.
@@ -13598,6 +14113,11 @@ interface GlassdoorJobsInput {
      * One of: 24h, week, month.
      */
     postedLimit?: "24h" | "week" | "month";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Job title or keywords to search (keyword mode). Provide this or a url.
      */
@@ -13714,6 +14234,11 @@ declare class GlassdoorNamespace {
  * Input for Google AI Mode (google.ai_mode).
  */
 interface GoogleAiModeInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The question or prompt to answer with Google AI Mode.
      */
@@ -13755,6 +14280,11 @@ interface GoogleAiModeData {
  * Input for Google AI Overview (google.ai_overview).
  */
 interface GoogleAiOverviewInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The question or prompt to answer with a Google AI Overview.
      */
@@ -13814,6 +14344,11 @@ interface GoogleAutocompleteInput {
      * Default: en.
      */
     hl?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The partial Google search query.
      */
@@ -13867,6 +14402,11 @@ interface GoogleImagesInput {
      * Fine-grained location for result localization, given as a canonical Google location string (e.g. 'New York, United States', 'London, United Kingdom'). More specific than the country-level gl.
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Image search query (e.g. golden gate bridge at sunset).
      */
@@ -13922,6 +14462,11 @@ interface GoogleImagesData {
  * Input for Google Lens (google.lens).
  */
 interface GoogleLensInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Public URL of the image to search with.
      * Format: uri.
@@ -13990,6 +14535,11 @@ interface GoogleNewsInput {
      * Fine-grained location for result localization, given as a canonical Google location string (e.g. 'New York, United States', 'London, United Kingdom'). More specific than the country-level gl.
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * News search query; supports operators like '-', 'OR', and 'site:' (e.g. bitcoin site:cnn.com).
      */
@@ -14037,6 +14587,11 @@ interface GoogleNewsData {
  * Input for Google Patents (google.patents).
  */
 interface GooglePatentsInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Google Patents search query.
      */
@@ -14113,6 +14668,11 @@ interface GooglePatentsData {
  * Input for Google Scholar (google.scholar).
  */
 interface GoogleScholarInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Google Scholar search query.
      */
@@ -14200,6 +14760,11 @@ interface GoogleSearchInput {
      * Fine-grained location for result localization, given as a canonical Google location string (e.g. 'New York, United States', 'London, United Kingdom'). More specific than the country-level gl.
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Google search query.
      */
@@ -14261,6 +14826,11 @@ interface GoogleSearch100Input {
      * Default: en.
      */
     hl?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Google search query.
      */
@@ -14328,6 +14898,11 @@ interface GoogleVideosInput {
      * Fine-grained location for result localization, given as a canonical Google location string (e.g. 'New York, United States', 'London, United Kingdom'). More specific than the country-level gl.
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The video search query.
      */
@@ -14520,6 +15095,11 @@ declare class GoogleNamespace {
  * Input for Google Ads Ad Details (google_ads.ad_details).
  */
 interface GoogleAdsAdDetailsInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Google Ads Transparency Center creative URL (e.g. "https://adstransparency.google.com/advertiser/AR.../creative/CR...").
      */
@@ -14568,6 +15148,11 @@ interface GoogleAdsAdDetailsData {
  * Input for Google Ads Advertiser Search (google_ads.advertiser_search).
  */
 interface GoogleAdsAdvertiserSearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Advertiser name or keyword to search for (e.g. "lululemon").
      */
@@ -14635,6 +15220,11 @@ interface GoogleAdsCompanyAdsInput {
      * One of: google_maps, google_play, google_search, google_shopping, youtube.
      */
     platform?: "google_maps" | "google_play" | "google_search" | "google_shopping" | "youtube";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Two-letter country code to scope results (e.g. "US", "AU").
      */
@@ -14707,6 +15297,11 @@ interface GoogleAdsSearchInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * A Google Ads Transparency Center URL for a selected advertiser or domain (e.g. https://adstransparency.google.com/advertiser/AR01614014350098432001?region=US).
      */
@@ -14844,6 +15439,11 @@ declare class GoogleAdsNamespace {
  * Input for Google Finance Quote (google_finance.quote).
  */
 interface GoogleFinanceQuoteInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The symbol to quote. US stocks use a plain ticker (e.g. AAPL, TSLA); non-US stocks add a market suffix (e.g. VOW3.DE, BABA.HK, BARC.L); indices use a caret (e.g. ^GSPC, ^DJI); crypto and currencies use pair form (e.g. BTC-USD, EURUSD=X); mutual funds and futures use their symbol (e.g. VFIAX, ES=F). Common alternate forms are accepted and normalized (e.g. AAPL:NASDAQ, .DJI, BTC/USD). Exact symbols only, not a company-name search.
      */
@@ -14992,6 +15592,11 @@ interface GoogleShoppingSearchInput {
      * Range: minimum 1, maximum 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Product name, brand, or keywords to search for (e.g. "Nike running shoes").
      */
@@ -15099,6 +15704,11 @@ interface HackernewsProfileInput {
      * Hacker News username, e.g. "pg".
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Hacker News Profile (hackernews.profile).
@@ -15116,6 +15726,11 @@ interface HackernewsProfileData {
  * Input for Hacker News Search (hackernews.search).
  */
 interface HackernewsSearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keyword, e.g. "ai".
      */
@@ -15173,6 +15788,11 @@ interface HackernewsStoryInput {
      * Hacker News story id, e.g. "47340079".
      */
     id: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Hacker News Story (hackernews.story).
@@ -15212,6 +15832,11 @@ interface HackernewsStoryCommentsInput {
      * Hacker News story id, e.g. "47340079".
      */
     id: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface HackernewsStoryCommentsComment {
     /**
@@ -15321,6 +15946,11 @@ interface IndeedJobsInput {
      * One of: 24h, week.
      */
     postedLimit?: "24h" | "week";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Job search keywords (e.g. software engineer).
      */
@@ -15407,6 +16037,11 @@ interface InstagramAudioReelsInput {
      * Pagination cursor returned by a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramAudioReelsReel {
     code: string;
@@ -15441,6 +16076,11 @@ interface InstagramAudioReelsData {
  * Input for Instagram Basic Profile (instagram.basic_profile).
  */
 interface InstagramBasicProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Instagram numeric user id.
      */
@@ -15486,6 +16126,11 @@ interface InstagramCommentRepliesInput {
      * Pagination cursor from a previous response's nextCursor.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Instagram post or reel URL the comment belongs to.
      */
@@ -15543,6 +16188,11 @@ interface InstagramEmbedInput {
      * Instagram username without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Instagram Profile Embed (instagram.embed).
@@ -15567,6 +16217,11 @@ interface InstagramFollowersInput {
      * Range: minimum 50, maximum 1000.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Set true if you intend to page through followers, so the request is only served by a source that can return a nextCursor. The bulk source cannot page, and a paging source may cost more per request. Cannot be combined with requireSinglePage.
      */
@@ -15637,6 +16292,11 @@ interface InstagramFollowingInput {
      * Range: minimum 50, maximum 1000.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Set true if you intend to page through the following list, so the request is only served by a source that can return a nextCursor. The bulk source cannot page, and a paging source may cost more per request. Cannot be combined with requireSinglePage.
      */
@@ -15707,6 +16367,11 @@ interface InstagramHashtagAnalyticsInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramHashtagAnalyticsItem {
     difficulty?: string;
@@ -15755,6 +16420,11 @@ interface InstagramHashtagRecentPostsInput {
      * Hashtag to monitor, without the leading #.
      */
     hashtag: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramHashtagRecentPostsPost {
     /**
@@ -15854,6 +16524,11 @@ interface InstagramHashtagTopPostsInput {
      * Hashtag to fetch, without the leading #.
      */
     hashtag: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramHashtagTopPostsPost {
     /**
@@ -15950,6 +16625,11 @@ interface InstagramHighlightDetailInput {
      * The id of the highlight to retrieve details for.
      */
     id: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Instagram Highlight Detail (instagram.highlight_detail).
@@ -15985,6 +16665,11 @@ interface InstagramHighlightDetailData {
  * Input for Instagram Media Transcript (instagram.media_transcript).
  */
 interface InstagramMediaTranscriptInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Instagram post or reel URL.
      */
@@ -16018,6 +16703,11 @@ interface InstagramMediaTranscriptData {
  * Input for Instagram Post (instagram.post).
  */
 interface InstagramPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Set true to be served only by a source that reports a reel's play count. The default cheapest source does not carry play counts, so `plays` is absent from its responses; opting in guarantees the field when Instagram exposes it, at a higher price per request.
      */
@@ -16061,6 +16751,11 @@ interface InstagramPostCommentsInput {
      * Pagination cursor from a previous response's nextCursor.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Instagram post or reel URL.
      */
@@ -16112,6 +16807,11 @@ interface InstagramProfileInput {
      * Instagram username without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramProfileBioLink {
     /**
@@ -16198,6 +16898,11 @@ interface InstagramProfileData {
  * Input for Instagram Reel Transcript (instagram.reel_transcript).
  */
 interface InstagramReelTranscriptInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The URL of a public Instagram reel or video post with spoken audio (e.g. https://www.instagram.com/reel/C8yKXdRxKqK/).
      */
@@ -16298,6 +17003,11 @@ interface InstagramReelsSearchInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keyword (e.g. "crossfit").
      */
@@ -16368,6 +17078,11 @@ interface InstagramSearchInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search Instagram for; one or more words without special punctuation (e.g. coffee roastery).
      */
@@ -16453,6 +17168,11 @@ interface InstagramSearchHashtagInput {
      * One of: all, reel.
      */
     mediaType?: "all" | "reel";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramSearchHashtagPost {
     /**
@@ -16541,6 +17261,11 @@ interface InstagramSearchProfilesInput {
      * Pagination cursor returned by a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Bio or caption keyword/phrase to search for.
      */
@@ -16591,6 +17316,11 @@ interface InstagramSearchProfilesData {
  * Input for Instagram Stories (full) (instagram.stories_full).
  */
 interface InstagramStoriesFullInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Instagram username or handle without the @.
      */
@@ -16661,6 +17391,11 @@ interface InstagramStoriesFullData {
  * Input for Instagram Stories (basic) (instagram.stories_thin).
  */
 interface InstagramStoriesThinInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Instagram username/handle to fetch currently live stories for (without the @).
      */
@@ -16713,6 +17448,11 @@ interface InstagramTaggedPostsInput {
      * Instagram username without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramTaggedPostsPost {
     /**
@@ -16767,6 +17507,11 @@ interface InstagramTaggedPostsData {
  * Input for Instagram Trending Reels (instagram.trending_reels).
  */
 interface InstagramTrendingReelsInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramTrendingReelsReel {
     caption: string;
@@ -16807,6 +17552,11 @@ interface InstagramUserHighlightsInput {
      * Instagram username without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Instagram numeric user id (optional, faster than handle).
      */
@@ -16852,6 +17602,11 @@ interface InstagramUserPostsInput {
      * Instagram username without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface InstagramUserPostsPost {
     /**
@@ -16940,6 +17695,11 @@ interface InstagramUserReelsInput {
      * Instagram handle.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Instagram user id (faster than handle when known).
      */
@@ -17774,6 +18534,11 @@ interface JobSearchTheirstackInput {
      * Date posted max age in days. If 0, only return jobs posted today. If 1, from today and yesterday, etc.
      */
     postedAtMaxAgeDays?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     propertyExistsAnd?: string[];
     /**
      * Return jobs that have any of these fields not null. For example, if you pass ['final_url'], it will return jobs that have a final_url set. This field also support chaining of fields. For example, if you pass ['company_object.domain', 'company_object.linkedin_url'], it will return jobs that have a company domain or a company linkedin_url set.
@@ -18298,6 +19063,11 @@ declare class JobSearchNamespace {
  * Input for LinkedIn Ad Details (linkedin.ad).
  */
 interface LinkedinAdInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * LinkedIn Ad Library ad URL (e.g. "https://www.linkedin.com/ad-library/detail/666281156").
      */
@@ -18354,6 +19124,11 @@ interface LinkedinAdsInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * LinkedIn Ad Library search URL or a LinkedIn company URL (e.g. https://www.linkedin.com/ad-library/search?companyIds=1035).
      */
@@ -18431,6 +19206,11 @@ interface LinkedinAdsSearchInput {
      * Opaque pagination token from a previous response's nextCursor.
      */
     paginationToken?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search start date in YYYY-MM-DD format.
      */
@@ -18474,6 +19254,11 @@ interface LinkedinAdsSearchData {
  * Input for LinkedIn Article (linkedin.article).
  */
 interface LinkedinArticleInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Public LinkedIn article or newsletter issue URL, e.g. https://www.linkedin.com/pulse/your-article-slug. Pair it with the attachmentUrl returned by linkedin.search_posts_full to read the article behind a post.
      * Format: uri.
@@ -18543,6 +19328,11 @@ interface LinkedinArticleData {
  * Input for LinkedIn Company (linkedin.company).
  */
 interface LinkedinCompanyInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn company page URL.
      */
@@ -18689,6 +19479,11 @@ interface LinkedinCompanyEmployeesInput {
      * Range: minimum 1, maximum 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * An employee: name, headline, location, handle, and LinkedIn profile URL.
@@ -18762,6 +19557,11 @@ interface LinkedinCompanyPostsInput {
      * One of: any, 1h, 24h, week, month, 3months, 6months, year.
      */
     postedLimit?: "any" | "1h" | "24h" | "week" | "month" | "3months" | "6months" | "year";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn company page URL.
      */
@@ -18883,6 +19683,11 @@ interface LinkedinCompanyPostsThinInput {
      * Range: minimum 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn company page URL.
      */
@@ -18920,6 +19725,11 @@ interface LinkedinCompanyPostsThinData {
  * Input for LinkedIn Company (basic) (linkedin.company_thin).
  */
 interface LinkedinCompanyThinInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn company page URL.
      */
@@ -18963,6 +19773,11 @@ interface LinkedinCompanyThinData {
  * Input for LinkedIn Email Finder (linkedin.email).
  */
 interface LinkedinEmailInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * LinkedIn profile URL or public identifier (the last part of the URL) to find the deliverability-validated work email for.
      */
@@ -19060,6 +19875,11 @@ interface LinkedinJobsInput {
      * One of: 1h, 24h, week, month.
      */
     postedLimit?: "1h" | "24h" | "week" | "month";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Job title or keywords to search. Supports LinkedIn boolean operators.
      */
@@ -19231,6 +20051,11 @@ interface LinkedinJobsThinInput {
      * One of: 24h, week, month.
      */
     postedLimit?: "24h" | "week" | "month";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Job title or keywords to search.
      */
@@ -19294,6 +20119,11 @@ interface LinkedinJobsThinData {
  * Input for LinkedIn Post (linkedin.post).
  */
 interface LinkedinPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn post or article URL.
      */
@@ -19348,6 +20178,11 @@ interface LinkedinPostCommentsInput {
      * One of: any, 24h, week, month, 3months, 6months, year.
      */
     postedLimit?: "any" | "24h" | "week" | "month" | "3months" | "6months" | "year";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full URL of the LinkedIn post to list comments for.
      */
@@ -19442,6 +20277,11 @@ interface LinkedinPostReactionsInput {
      * Range: minimum 1, maximum 100.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * URL of the LinkedIn post to list reactions for (a /posts/...-activity-... or /feed/update/urn:li:activity:... link).
      */
@@ -19501,6 +20341,11 @@ interface LinkedinPostReactionsData {
  * Input for LinkedIn Post Transcript (linkedin.post_transcript).
  */
 interface LinkedinPostTranscriptInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The full URL of the LinkedIn post to get the video transcript from.
      */
@@ -19522,6 +20367,11 @@ interface LinkedinPostTranscriptData {
  * Input for LinkedIn Profile (linkedin.profile).
  */
 interface LinkedinProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn profile URL.
      */
@@ -19806,6 +20656,11 @@ interface LinkedinProfilePostsFullInput {
      * Only return posts published on or after this date or timestamp, using a JavaScript-compatible date-time string.
      */
     postedLimitDate?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full URL of the public LinkedIn profile whose posts should be returned.
      * Format: uri.
@@ -20273,6 +21128,11 @@ interface LinkedinProfilePostsThinInput {
      * Default: 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full URL of the public LinkedIn profile whose posts should be returned.
      * Format: uri.
@@ -20430,6 +21290,11 @@ interface LinkedinProfilePostsThinData {
  * Input for LinkedIn Profile (basic) (linkedin.profile_thin).
  */
 interface LinkedinProfileThinInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full LinkedIn profile URL.
      */
@@ -20567,6 +21432,11 @@ interface LinkedinSearchCompaniesInput {
      * Optional location filter, written out in full (e.g. United Kingdom or San Francisco).
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search LinkedIn companies for (e.g. marketing agency).
      */
@@ -20642,6 +21512,11 @@ interface LinkedinSearchPostsInput {
      * One of: last-hour, last-day, last-week, last-month, last-year.
      */
     datePosted?: "last-hour" | "last-day" | "last-week" | "last-month" | "last-year";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The post search query.
      */
@@ -20744,6 +21619,11 @@ interface LinkedinSearchPostsFullInput {
      * Only return posts mentioning these LinkedIn member profile URLs.
      */
     mentioningMemberUrls?: string[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * LinkedIn post search query, including quoted terms or Boolean operators accepted by LinkedIn search.
      */
@@ -20996,6 +21876,11 @@ interface LinkedinSearchProfilesInput {
      * Filter by a past job title the person held (e.g. ['Product Manager']).
      */
     pastJobTitles?: string[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Filter by the profile's primary language.
      */
@@ -21206,6 +22091,11 @@ interface LinkedinSearchProfilesEmailInput {
      * Filter by a past job title the person held (e.g. ['Product Manager']).
      */
     pastJobTitles?: string[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Filter by the profile's primary language.
      */
@@ -21343,6 +22233,11 @@ interface LinkedinSearchProfilesEmailData {
  * Input for LinkedIn Profile Search (basic) (linkedin.search_profiles_thin).
  */
 interface LinkedinSearchProfilesThinInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search query for LinkedIn profiles - a role, name, or keywords (e.g. 'Marketing Manager').
      */
@@ -21722,6 +22617,11 @@ interface MapsContactsInput {
      * One of: two, twoAndHalf, three, threeAndHalf, four, fourAndHalf.
      */
     placeMinimumStars?: "two" | "twoAndHalf" | "three" | "threeAndHalf" | "four" | "fourAndHalf";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * What you would type in the Google Maps search bar (e.g. dentist).
      */
@@ -21866,6 +22766,11 @@ interface MapsPlaceInput {
      * One of: two, twoAndHalf, three, threeAndHalf, four, fourAndHalf.
      */
     placeMinimumStars?: "two" | "twoAndHalf" | "three" | "threeAndHalf" | "four" | "fourAndHalf";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The business name or search text to look up, as you would type it into the Google Maps search bar (e.g. Blue Bottle Coffee).
      */
@@ -21998,6 +22903,11 @@ interface MapsReviewsInput {
      * One of: 24h, week, month, year.
      */
     postedLimit?: "24h" | "week" | "month" | "year";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only return reviews whose text contains this keyword or phrase (case-insensitive). Omit to return all reviews (e.g. parking).
      */
@@ -22112,6 +23022,11 @@ interface MapsSearchInput {
      * One of: two, twoAndHalf, three, threeAndHalf, four, fourAndHalf.
      */
     placeMinimumStars?: "two" | "twoAndHalf" | "three" | "threeAndHalf" | "four" | "fourAndHalf";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * What you would type in the Google Maps search bar (e.g. coffee shop).
      */
@@ -22284,6 +23199,11 @@ interface MobilePhoneAiArkInput {
      * Format: uri.
      */
     linkedinUrl?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Mobile Phone - AI Ark (mobile_phone.ai_ark).
@@ -22324,6 +23244,11 @@ interface MobilePhoneLeadmagicInput {
      * Format: email.
      */
     personalEmail?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Format: uri.
      */
@@ -22396,6 +23321,11 @@ interface NaverBlogSearchInput {
      * Range: minimum 1, maximum 5.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword phrase to search across Naver blogs.
      */
@@ -22504,6 +23434,11 @@ interface PandaexpressLocationsInput {
      * Range: minimum -180, maximum 180.
      */
     longitude: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search radius in miles (default 10).
      * Range: minimum 1, maximum 100.
@@ -22566,6 +23501,11 @@ interface PandaexpressLocationsData {
  * Input for Panda Express Menu (pandaexpress.menu).
  */
 interface PandaexpressMenuInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Panda Express restaurant id (the `id` from Panda Express Locations).
      */
@@ -22611,6 +23551,11 @@ interface PandaexpressMenuData {
  * Input for Panda Express Nutrition (pandaexpress.nutrition).
  */
 interface PandaexpressNutritionInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Menu item name (or substring) to look up, e.g. "orange chicken" or "chow mein".
      */
@@ -22707,6 +23652,11 @@ interface PeopleSearchAiArkInput {
      * Default: 0.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Maximum people to return on this page.
      * Range: minimum 1, maximum 100.
@@ -23340,6 +24290,11 @@ interface PeopleSearchCrustdataV3Input {
      * Default: 3.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     profileKeywords?: unknown;
     /**
      * Default: false.
@@ -24300,6 +25255,11 @@ interface PeopleSearchFullenrichInput {
      * Filter by a skill the person lists. Each entry is an object taking a `value` string and an optional `exact_match` boolean, e.g. [{"value": "stripe.com", "exact_match": true}]. A bare string is rejected.
      */
     personSkills?: PeopleSearchFullenrichPersonSkill[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface PeopleSearchFullenrichPeople {
     /**
@@ -24598,6 +25558,11 @@ interface PeopleSearchLushaInput {
          */
         size?: number;
     };
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface PeopleSearchLushaPeople {
     /**
@@ -24763,6 +25728,11 @@ interface PeopleSearchPeopledatalabsInput {
      * Default: 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Elasticsearch-style query over the People Data Labs person dataset, e.g. {"bool": {"must": [{"term": {"job_company_website": "posthog.com"}}]}}. Send this or sql, never both.
      */
@@ -25358,6 +26328,11 @@ interface PeopleSearchProspeoInput {
      * Filter by total years of experience, as a numeric range.
      */
     personYearOfExperience?: {};
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface PeopleSearchProspeoPeople {
     /**
@@ -25990,6 +26965,11 @@ interface PeopleSearchQuickenrichInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Filter on employer revenue band.
      */
@@ -26172,6 +27152,11 @@ interface PeopleSearchQuickenrichCompanyInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * One job title, or several comma-separated, e.g. "CEO, CFO".
      */
@@ -26397,6 +27382,11 @@ declare class PeopleSearchNamespace {
  * Input for Perplexity Search (perplexity.search).
  */
 interface PerplexitySearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Question or research prompt for Perplexity to answer using web search.
      */
@@ -26475,6 +27465,11 @@ interface PersonSkipTraceInput {
      * Phone number to reverse-trace (e.g. 415-555-2671).
      */
     phone?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * A skip-trace match: name, age, current and past addresses, phone numbers, email addresses, relatives, and associates.
@@ -26619,6 +27614,11 @@ interface PersonEnrichmentAviatoInput {
      */
     linkedinURL?: string;
     polyworkID?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     require?: string[];
     signalNfxID?: string;
     twitterID?: string;
@@ -27317,6 +28317,11 @@ interface PersonEnrichmentBettercontactInput {
      * Format: uri.
      */
     linkedinUrl?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Person Enrichment - BetterContact (person_enrichment.bettercontact).
@@ -27560,6 +28565,11 @@ interface PersonEnrichmentFullenrichBulkInput {
      * People to enrich, up to 99 per call. You are charged only for the contacts the waterfall resolves, though the funds held cover every contact you submit until the call settles. Each entry is passed to FullEnrich exactly as you write it, which is why these keys are snake_case while the rest of the API is camelCase. Give a name plus an employer (company_name or domain), or a linkedin_url, or both - more identity means a better hit rate.
      */
     contacts: PersonEnrichmentFullenrichBulkContact[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface PersonEnrichmentFullenrichBulkContact {
     /**
@@ -27932,6 +28942,11 @@ interface PersonEnrichmentFullenrichReverseEmailInput {
      * Addresses to look up, up to 99 per call. You are charged only for the addresses that resolve to a person, though the funds held cover every address you submit until the call settles. Each entry is an object so you can tag it; the address itself goes in email.
      */
     contacts: PersonEnrichmentFullenrichReverseEmailContact[];
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface PersonEnrichmentFullenrichReverseEmailContact {
     /**
@@ -28252,6 +29267,11 @@ interface PersonEnrichmentLushaInput {
      * Format: uri.
      */
     linkedinUrl?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Force a fresh job title and employer lookup instead of the cached one.
      */
@@ -28609,6 +29629,11 @@ interface PersonEnrichmentPeopledatalabsInput {
      * Postal or ZIP code to match on.
      */
     postalCode?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Social profile URL the person has used, e.g. a LinkedIn, Twitter, Facebook or GitHub profile. The strongest single identifier.
      */
@@ -29226,6 +30251,11 @@ interface PersonEnrichmentProspeoInput {
      * Prospeo person id from an earlier People Search - Prospeo call.
      */
     personId?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface PersonEnrichmentProspeoEvent {
     /**
@@ -29721,6 +30751,11 @@ interface PersonEnrichmentQuickenrichInput {
      * Format: email.
      */
     email: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Person Enrichment - QuickEnrich (person_enrichment.quickenrich).
@@ -29918,6 +30953,11 @@ interface PinterestSearchInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword, topic, brand, or theme to search Pinterest for (e.g. mid-century living room).
      */
@@ -30001,6 +31041,11 @@ interface PlaystoreReviewsInput {
      * Range: minimum 1, maximum 100.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only return reviews with this exact star rating from 1 to 5 (e.g. 1); omit for all ratings.
      */
@@ -30091,6 +31136,11 @@ interface PolymarketMarketsInput {
      * Range: minimum 1, maximum 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search term for markets (e.g. election, bitcoin, super bowl).
      */
@@ -30222,6 +31272,11 @@ interface RealtorSearchInput {
      * City, ZIP code, neighborhood or state to search (e.g. Las Vegas, NV).
      */
     location: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Maximum listing price in USD (e.g. 750000).
      * Range: minimum 0.
@@ -30372,6 +31427,11 @@ declare class RealtorNamespace {
  * Input for Reddit Post (reddit.post).
  */
 interface RedditPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Reddit post URL in the /r/<subreddit>/comments/<id>/<slug>/ form, e.g. "https://www.reddit.com/r/IAmA/comments/z1c9z/i_am_barack_obama_president_of_the_united_states/". The short "reddit.com/comments/<id>" form is not accepted.
      */
@@ -30444,6 +31504,11 @@ interface RedditPostCommentsInput {
      * Cursor from a previous response for more comments.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Reddit post URL.
      */
@@ -30516,6 +31581,11 @@ interface RedditPostTranscriptInput {
      * Optional two-letter language code (defaults to en).
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Reddit post URL or direct v.redd.it video URL to transcribe.
      */
@@ -30541,6 +31611,11 @@ interface RedditPostTranscriptData {
  * Input for Reddit Profile (reddit.profile).
  */
 interface RedditProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Reddit username, without the u/ prefix. Example: "spez".
      */
@@ -30641,6 +31716,11 @@ interface RedditSearchInput {
      * Opaque pagination cursor from a previous response's nextCursor. Omit for the first page; pass it to fetch the next page of results.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Free-text search across all of Reddit. Reddit's field operators are supported inside the string: subreddit:<name> to scope to one subreddit, author:<user>, title:<text>, selftext:<text>, self:yes|no, nsfw:yes|no, and boolean AND/OR/NOT. To restrict to a single subreddit you can use subreddit:<name> here, or use the reddit.subreddit_posts SKU for a plain subreddit listing.
      */
@@ -30735,6 +31815,11 @@ interface RedditSearchData {
  * Input for Reddit Subreddit Details (reddit.subreddit_details).
  */
 interface RedditSubredditDetailsInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Subreddit name without the r/ prefix. Case-sensitive (e.g. "AskReddit", not "askreddit").
      */
@@ -30792,6 +31877,11 @@ interface RedditSubredditPostsInput {
      * Default: 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Listing sort order.
      * One of: hot, new, top.
@@ -30868,6 +31958,11 @@ interface RedditSubredditSearchInput {
      * Optional pagination token from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Optional search query to match posts (e.g. 'push ups').
      */
@@ -30955,6 +32050,11 @@ interface RedditTrendingPostsInput {
      * Default: 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface RedditTrendingPostsPost {
     /**
@@ -31022,6 +32122,11 @@ interface RedditUserCommentsInput {
      * Default: 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Sort order for the user's comments. Defaults to new (most recent first).
      * One of: new, top, hot, controversial.
@@ -31090,6 +32195,11 @@ interface RedditUserPostsInput {
      * Opaque pagination cursor from a previous response's nextCursor. Omit for the first page; pass it back to fetch the next page.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Sort order for the user's posts. Defaults to new (most recent first).
      * One of: new, top, hot, controversial.
@@ -31358,6 +32468,11 @@ interface RedfinSearchInput {
      * Range: minimum 1, maximum 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Redfin search results URL for a city, ZIP or map area (e.g. https://www.redfin.com/city/30772/CA/San-Francisco).
      */
@@ -31483,6 +32598,11 @@ interface RednoteNoteInput {
      * RedNote (Xiaohongshu) note ID.
      */
     noteId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of RedNote (Xiaohongshu) Note (rednote.note).
@@ -31582,6 +32702,11 @@ interface RednoteNoteCommentsInput {
      * RedNote (Xiaohongshu) note ID.
      */
     noteId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface RednoteNoteCommentsComment {
     /**
@@ -31656,6 +32781,11 @@ interface RednoteNoteCommentsData {
  * Input for RedNote (Xiaohongshu) Profile (rednote.profile).
  */
 interface RednoteProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * RedNote (Xiaohongshu) user ID.
      */
@@ -31717,6 +32847,11 @@ interface RednoteSearchInput {
      * Pagination cursor from the previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search for on RedNote (Xiaohongshu).
      */
@@ -31822,6 +32957,11 @@ interface RednoteSearchUsersInput {
      * Pagination cursor from the previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search for on RedNote (Xiaohongshu).
      */
@@ -31888,6 +33028,11 @@ interface RednoteUserNotesInput {
      * Pagination cursor from the previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * RedNote (Xiaohongshu) user ID.
      */
@@ -32097,6 +33242,11 @@ interface SecFilingsInput {
      * Range: minimum 1, maximum 25.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Company stock ticker symbol, e.g. AAPL, MSFT, or TSLA. Provide either ticker or companyName; ticker is the more precise lookup.
      */
@@ -32191,6 +33341,11 @@ interface SemrushKeywordsInput {
      * The search term to research (e.g. "best running shoes").
      */
     keyword: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SemrushKeywordsItem {
     /**
@@ -32299,6 +33454,11 @@ interface SemrushOverviewInput {
      * Default: false.
      */
     includeMoz?: boolean;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SemrushOverviewItem {
     /**
@@ -32484,6 +33644,11 @@ interface SeoCompetitorsDomainInput {
      * One of: intersections_desc, organic_keywords_desc, organic_etv_desc, avg_position_asc.
      */
     orderBy?: "intersections_desc" | "organic_keywords_desc" | "organic_etv_desc" | "avg_position_asc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Domain to analyze, without a protocol or leading www.
      */
@@ -32556,6 +33721,11 @@ interface SeoDomainIntersectionInput {
      * One of: volume_desc, volume_asc, cpc_desc, difficulty_asc, difficulty_desc.
      */
     orderBy?: "volume_desc" | "volume_asc" | "cpc_desc" | "difficulty_asc" | "difficulty_desc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * First domain to compare, without a protocol or leading www.
      */
@@ -32627,6 +33797,11 @@ interface SeoDomainRankOverviewInput {
      * Default: 2840.
      */
     location?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Domain to analyze, without a protocol or leading www.
      */
@@ -32712,6 +33887,11 @@ interface SeoDomainTechnologiesInput {
      * Domain to analyze, without a protocol or leading www. Billing is flat per request.
      */
     domain: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of SEO Domain Technologies (seo.domain_technologies).
@@ -32800,6 +33980,11 @@ interface SeoDomainsByTechnologyInput {
      * One of: rank_desc, rank_asc, last_visited_desc, country_asc.
      */
     orderBy?: "rank_desc" | "rank_asc" | "last_visited_desc" | "country_asc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Exact technology name to match, for example Nginx or HubSpot. Only technologies present in the detection index are accepted; an unindexed name returns found false, so use keyword instead when a product is not matched by name.
      */
@@ -32893,6 +34078,11 @@ interface SeoKeywordDifficultyInput {
      * Default: 2840.
      */
     location?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoKeywordDifficultyDifficultie {
     /**
@@ -32948,6 +34138,11 @@ interface SeoKeywordIdeasInput {
      * One of: volume_desc, volume_asc, cpc_desc, difficulty_asc, difficulty_desc.
      */
     orderBy?: "volume_desc" | "volume_asc" | "cpc_desc" | "difficulty_asc" | "difficulty_desc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoKeywordIdeasIdea {
     /**
@@ -33012,6 +34207,11 @@ interface SeoKeywordOverviewInput {
      * Default: 2840.
      */
     location?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoKeywordOverviewKeyword {
     /**
@@ -33118,6 +34318,11 @@ interface SeoKeywordSuggestionsInput {
      * One of: volume_desc, volume_asc, cpc_desc, difficulty_asc, difficulty_desc.
      */
     orderBy?: "volume_desc" | "volume_asc" | "cpc_desc" | "difficulty_asc" | "difficulty_desc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoKeywordSuggestionsSuggestion {
     /**
@@ -33191,6 +34396,11 @@ interface SeoLocalPackInput {
      * Precise geo target as latitude,longitude or latitude,longitude,radius (radius in meters); for example, 40.7580,-73.9855 or 40.7580,-73.9855,1000. Supply either location or locationCoordinate, not both.
      */
     locationCoordinate?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoLocalPackPlace {
     /**
@@ -33279,6 +34489,11 @@ interface SeoRankedKeywordsInput {
      * One of: position_asc, position_desc, volume_desc, etv_desc.
      */
     orderBy?: "position_asc" | "position_desc" | "volume_desc" | "etv_desc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Domain to analyze, without a protocol or leading www.
      */
@@ -33372,6 +34587,11 @@ interface SeoRelatedKeywordsInput {
      * One of: volume_desc, volume_asc, cpc_desc, difficulty_asc, difficulty_desc.
      */
     orderBy?: "volume_desc" | "volume_asc" | "cpc_desc" | "difficulty_asc" | "difficulty_desc";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoRelatedKeywordsRelatedKeyword {
     /**
@@ -33436,6 +34656,11 @@ interface SeoSearchIntentInput {
      * Default: en.
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SeoSearchIntentIntent {
     /**
@@ -33483,6 +34708,11 @@ interface SeoSearchVolumeInput {
      * Default: 2840.
      */
     location?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * When true, include Google search-partner network volume in the reported numbers; when false (the default), count Google search only.
      */
@@ -33724,6 +34954,11 @@ declare class SeoNamespace {
  * Input for Snapchat Profile (snapchat.profile).
  */
 interface SnapchatProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Snapchat username or profile URL to look up (e.g. fcbarcelona or https://www.snapchat.com/add/fcbarcelona).
      */
@@ -33829,6 +35064,11 @@ interface SocialFinderInput {
      * Limit the search to one network: askfm, discord, facebook, github, instagram, linkedin, medium, pinterest, steam, threads, tiktok, twitch, or youtube (e.g. instagram); all networks are searched when omitted.
      */
     platform?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface SocialFinderItem {
     /**
@@ -33883,6 +35123,11 @@ interface SpotifyAlbumInput {
      * Spotify album ID (alternative to url).
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Spotify album URL (e.g. https://open.spotify.com/album/0pgrg7phBbnwGJ2HBEl9EG).
      */
@@ -33937,6 +35182,11 @@ interface SpotifyArtistInput {
      * Spotify artist ID (alternative to url).
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Spotify artist URL (e.g. https://open.spotify.com/artist/3DiDSECUqqY1AuBP8qtaIa).
      */
@@ -33990,6 +35240,11 @@ interface SpotifyArtistData {
  * Input for Spotify Play Count (spotify.play_count).
  */
 interface SpotifyPlayCountInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Spotify track, album, or artist URL to fetch stream counts for (e.g. https://open.spotify.com/track/3n3Ppam7vgaVa1iaRUc9Lp).
      */
@@ -34047,6 +35302,11 @@ interface SpotifyPodcastInput {
      * Spotify podcast show ID (alternative to url).
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Spotify podcast show URL (e.g. https://open.spotify.com/show/3mliji9352UAk3XnWElnDV).
      */
@@ -34092,6 +35352,11 @@ interface SpotifyPodcastEpisodesInput {
      * Spotify podcast show ID (alternative to url).
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Spotify podcast show URL (e.g. https://open.spotify.com/show/4rOoJ6Egrf8K2IrywzwOMk).
      */
@@ -34139,6 +35404,11 @@ interface SpotifyPodcastEpisodesData {
  * Input for Spotify Search (spotify.search).
  */
 interface SpotifySearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search term (e.g. "my first million").
      */
@@ -34208,6 +35478,11 @@ interface SpotifyTrackInput {
      * Spotify track ID (alternative to url).
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Spotify track URL (e.g. https://open.spotify.com/track/4cOdK2wGLETKBW3PvgPWqT).
      */
@@ -34367,6 +35642,11 @@ interface SubstackPostsInput {
      * Default: false.
      */
     onlyFree?: boolean;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only return posts published on or after this date, format YYYY-MM-DD (e.g. 2024-01-01). Applied within the most recent 'limit' posts scanned, so raise 'limit' to reach older date ranges.
      */
@@ -34574,6 +35854,11 @@ interface TechnographicsTheirstackInput {
      * Page number. Required when using page-based pagination.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Deprecated: use `keyword_category_slug_or` instead. Will return companies that have mentioned any keyword from any of these categories in their jobs. Case sensitive. Pass slugs.
      */
@@ -34718,6 +36003,11 @@ declare class TechnographicsNamespace {
  * Input for Threads Post (threads.post).
  */
 interface ThreadsPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The full URL of the Threads post to fetch (e.g. https://www.threads.com/@zuck/post/C8yKXdRxKqK).
      */
@@ -34773,6 +36063,11 @@ interface ThreadsPostData {
  * Input for Threads Profile (threads.profile).
  */
 interface ThreadsProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The Threads username to look up, without the @ prefix (e.g. zuck).
      */
@@ -34811,6 +36106,11 @@ interface ThreadsProfileData {
  * Input for Threads Search (threads.search).
  */
 interface ThreadsSearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword or hashtag to search public Threads posts for; the # prefix is optional (e.g. AI agents).
      */
@@ -34872,6 +36172,11 @@ interface ThreadsSearchData {
  * Input for Threads User Search (threads.search_users).
  */
 interface ThreadsSearchUsersInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The name or username to search Threads users for.
      */
@@ -34914,6 +36219,11 @@ interface ThreadsUserPostsInput {
      * The Threads username to list posts for, without the @ prefix.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface ThreadsUserPostsPost {
     /**
@@ -35039,6 +36349,11 @@ interface TiktokAdLibraryAdInput {
      * TikTok Top Ads material/ad ID, or a Top Ads detail URL (e.g. 7648493525660270600).
      */
     adId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of TikTok Ad Library Ad (tiktok.ad_library_ad).
@@ -35134,6 +36449,11 @@ interface TiktokAdLibrarySearchInput {
      * Time window for top ads. Use the canonical JSON integer 7, 30, or 180; legacy numeric strings remain accepted.
      */
     period?: unknown;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search ad titles and content (e.g. spotify).
      */
@@ -35219,6 +36539,11 @@ interface TiktokAdTransparencySearchInput {
      * Default: 0.
      */
     offset?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search in TikTok's EU Commercial Content Library. Provide query or advertiserId.
      */
@@ -35315,6 +36640,11 @@ interface TiktokAudienceDemographicsInput {
      * TikTok username without the leading @ (e.g. "shakira").
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TiktokAudienceDemographicsAudienceLocation {
     count: number;
@@ -35350,6 +36680,11 @@ interface TiktokCommentRepliesInput {
      * Pagination cursor from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * TikTok video URL the comment belongs to.
      */
@@ -35397,6 +36732,11 @@ interface TiktokFollowersInput {
      * TikTok username whose followers to list, without the @ prefix (e.g. stoolpresidente).
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TiktokFollowersFollower {
     avatarUrl: string;
@@ -35443,6 +36783,11 @@ interface TiktokFollowingInput {
      * TikTok username without the leading @ (e.g. "stoolpresidente").
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Set true if you intend to page through the following list, so the request is only served by a source that can return a nextCursor. Not all sources for this list can page, and one that can may cost more per request.
      */
@@ -35493,6 +36838,11 @@ interface TiktokHashtagVideosInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TiktokHashtagVideosItem {
     /**
@@ -35554,6 +36904,11 @@ interface TiktokLiveInput {
      * TikTok username without the leading @ (e.g. "thejustalex").
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of TikTok Live (tiktok.live).
@@ -35589,6 +36944,11 @@ interface TiktokLiveData {
  * Input for TikTok Photos (tiktok.photos).
  */
 interface TiktokPhotosInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full TikTok photo-mode post URL. TikTok serves slideshow posts under the same /video/<id> path as videos, so the normal share link works.
      */
@@ -35636,6 +36996,11 @@ interface TiktokProfileInput {
      * TikTok username without the leading @ (e.g. "stoolpresidente").
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of TikTok Profile (tiktok.profile).
@@ -35681,6 +37046,11 @@ interface TiktokProfileRegionInput {
      * TikTok username without the leading @ (e.g. "stoolpresidente").
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of TikTok Profile Region (tiktok.profile_region).
@@ -35712,6 +37082,11 @@ interface TiktokProfileVideosInput {
      * TikTok username without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TiktokProfileVideosVideo {
     /**
@@ -35757,6 +37132,11 @@ interface TiktokSearchHashtagInput {
      * Pagination cursor from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Hashtag or keyword to search for (without the leading #).
      */
@@ -35810,6 +37190,11 @@ interface TiktokSearchKeywordInput {
      * Time frame filter. Use a canonical JSON integer that is nonnegative; common values are 0 for any time, 1 for the past 24 hours, 7 for the past week, and 30 for the past month. Legacy numeric strings remain accepted.
      */
     datePosted?: unknown;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The keyword to search TikTok for.
      */
@@ -35860,6 +37245,11 @@ interface TiktokSearchTopInput {
      * Pagination cursor from a previous response.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Time-frame filter: yesterday, this-week, this-month, last-3-months, last-6-months, all-time.
      */
@@ -35922,6 +37312,11 @@ interface TiktokSearchUsersInput {
      * Pagination cursor from a previous response's nextCursor.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The keyword to search TikTok accounts for.
      */
@@ -35965,6 +37360,11 @@ interface TiktokSongInput {
      * The clip identifier for the song, found in TikTok music URLs (e.g. 7439295283975702544).
      */
     clipId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of TikTok Song (tiktok.song).
@@ -36014,6 +37414,11 @@ interface TiktokSongVideosInput {
      * Pagination cursor for retrieving the next page of results.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TiktokSongVideosVideo {
     /**
@@ -36095,6 +37500,11 @@ interface TiktokTopAdsSearchInput {
      * Default: 180.
      */
     period?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to search in TikTok Creative Center top video ads.
      */
@@ -36203,6 +37613,11 @@ interface TiktokTopAdsSearchData {
  * Input for TikTok Trending Feed (tiktok.trending_feed).
  */
 interface TiktokTrendingFeedInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * 2-letter country code for the proxy location (e.g. "US").
      */
@@ -36269,6 +37684,11 @@ interface TiktokTrendingHashtagsInput {
      * Default: 7.
      */
     period?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Two-letter country code of the market whose hashtag ranking to read (default US). Each market is ranked independently, so US and DE return different boards.
      * One of: US, FR, DE, IT, ES, GB, AR, AU, BR, CA, CO, EG, ID, IL, JP, KR, MY, MX, PH, SA, SG, ZA, TW, TH, TR, AE, VN.
@@ -36373,6 +37793,11 @@ interface TiktokTrendingHashtagsData {
  * Input for TikTok Video (tiktok.video).
  */
 interface TiktokVideoInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full TikTok video URL.
      */
@@ -36410,6 +37835,11 @@ interface TiktokVideoCommentsInput {
      * Pagination cursor from a previous response's nextCursor.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full TikTok video URL.
      */
@@ -36450,6 +37880,11 @@ interface TiktokVideoCommentsData {
  * Input for TikTok Video Download (tiktok.video_download).
  */
 interface TiktokVideoDownloadInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full TikTok video URL. Share links and tracking query params are fine.
      */
@@ -36496,6 +37931,11 @@ interface TiktokVideoDownloadData {
  * Input for TikTok Video Transcript (tiktok.video_transcript).
  */
 interface TiktokVideoTranscriptInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full TikTok video URL.
      */
@@ -36516,6 +37956,11 @@ interface TiktokVideoTranscriptData {
  * Input for TikTok Video Transcript (Audio) (tiktok.video_transcript_full).
  */
 interface TiktokVideoTranscriptFullInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * TikTok video URL (e.g. "https://www.tiktok.com/@user/video/1234567890").
      */
@@ -36998,6 +38443,11 @@ declare class TiktokNamespace {
  * Input for TikTok Shop Categories (tiktok_shop.categories).
  */
 interface TiktokShopCategoriesInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Country code of the TikTok Shop market whose category tree to list. Only US and VN publish a category tree.
      * One of: US, VN.
@@ -37068,6 +38518,11 @@ interface TiktokShopCategoryProductsInput {
      * TikTok Shop category id, from tiktok_shop.categories (e.g. 700645 for Health).
      */
     categoryId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Two-letter country code of the TikTok Shop market (e.g. US).
      * Default: US.
@@ -37151,6 +38606,11 @@ interface TiktokShopCreatorInput {
      * TikTok handle of the creator or shop account, without the @ (e.g. golinutrition).
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Lowercase two-letter country code of the TikTok Shop market (e.g. us).
      * Default: us.
@@ -37316,6 +38776,11 @@ interface TiktokShopCreatorData {
  * Input for TikTok Shop Product (tiktok_shop.product).
  */
 interface TiktokShopProductInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Two-letter country code for the proxy location used to access region-specific products (e.g. US, GB, FR). Defaults to US.
      */
@@ -37363,6 +38828,11 @@ interface TiktokShopProductData {
  * Input for TikTok Shop Product Full (tiktok_shop.product_full).
  */
 interface TiktokShopProductFullInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * TikTok Shop product URL. Any of the public forms works (https://www.tiktok.com/shop/pdp/<id>, https://shop.tiktok.com/<region>/pdp/<slug>/<id>, or https://shop.tiktok.com/view/product/<id>); the product id is read out of it.
      */
@@ -37512,6 +38982,11 @@ interface TiktokShopProductReviewsInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Two-letter country code of the product's shop region (e.g. US). Strongly recommended for correct results.
      */
@@ -37592,6 +39067,11 @@ interface TiktokShopSearchInput {
      * Range: minimum 1, maximum 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keyword for TikTok Shop products (e.g. wireless earbuds).
      */
@@ -37672,6 +39152,11 @@ interface TiktokShopSearchSuggestionsInput {
      * Default: en-US.
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Seed keyword to expand (e.g. ashwagandha gummies).
      */
@@ -37695,6 +39180,11 @@ interface TiktokShopShopProductsInput {
      * Opaque pagination cursor from a previous response's nextCursor.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Two-letter country code of the store's market (e.g. US).
      */
@@ -37764,6 +39254,11 @@ interface TiktokShopUserShowcaseInput {
      * The handle of the TikTok user (e.g. mrtiktokreviews).
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Geographical region for proxy placement (defaults to US).
      */
@@ -37953,6 +39448,11 @@ interface TripadvisorReviewsInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only return reviews whose bubble rating is in this set (e.g. ["5", "4"] for 4 and 5 star reviews); omit for all ratings.
      */
@@ -38030,6 +39530,11 @@ interface TripadvisorSearchInput {
      * Range: minimum 1, maximum 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Destination or keyword to search for (e.g. Barcelona).
      */
@@ -38183,6 +39688,11 @@ interface TrustpilotReviewsInput {
      * Range: minimum 1, maximum 200.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Review ordering: auto, relevancy, or recent (e.g. recent).
      * Default: auto.
@@ -38269,6 +39779,11 @@ declare class TrustpilotNamespace {
  * Input for Truth Social Post (truthsocial.post).
  */
 interface TruthsocialPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full Truth Social post URL, e.g. "https://truthsocial.com/@realDonaldTrump/posts/116824551176646175".
      */
@@ -38320,6 +39835,11 @@ interface TruthsocialProfileInput {
      * Truth Social handle without the @, e.g. "realDonaldTrump".
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Truth Social Profile (truthsocial.profile).
@@ -38368,6 +39888,11 @@ interface TruthsocialUserPostsInput {
      * Truth Social handle without the @, e.g. "realDonaldTrump".
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TruthsocialUserPostsPost {
     /**
@@ -38455,6 +39980,11 @@ declare class TruthsocialNamespace {
  * Input for X / Twitter Article (twitter.article).
  */
 interface TwitterArticleInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Canonical x.com or twitter.com URL of the public wrapper post for an X Article.
      */
@@ -38645,6 +40175,11 @@ interface TwitterArticleData {
  * Input for Twitter Community (twitter.community).
  */
 interface TwitterCommunityInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Community URL (e.g. https://x.com/i/communities/1926186499399139650).
      */
@@ -38688,6 +40223,11 @@ interface TwitterCommunityData {
  * Input for Twitter Community Tweets (twitter.community_tweets).
  */
 interface TwitterCommunityTweetsInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Community URL (e.g. https://x.com/i/communities/1926186499399139650).
      */
@@ -38739,6 +40279,11 @@ interface TwitterFollowersInput {
      * Default: 200.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Set true to get up to limit followers in one response instead of provider-native pages, served by a bulk provider when needed.
      */
@@ -38810,6 +40355,11 @@ interface TwitterFollowingInput {
      * Default: 200.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Set true to get up to limit accounts in one response instead of provider-native pages, served by a bulk provider when needed.
      */
@@ -38875,6 +40425,11 @@ interface TwitterProfileInput {
      * Twitter/X handle without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Twitter Profile (twitter.profile).
@@ -38912,6 +40467,11 @@ interface TwitterRepliesInput {
      * Range: minimum 1, maximum 40.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full URL of the X (Twitter) post to fetch replies for (e.g. https://x.com/nasa/status/1846987139428634858).
      */
@@ -39020,6 +40580,11 @@ interface TwitterSearchInput {
      * Default: 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search query using X (Twitter) advanced-search syntax. IMPORTANT: bare terms are ANDed - a tweet must contain EVERY word, so a list of loosely related keywords matches nothing; use one short phrase or OR between alternatives (e.g. 'anyapi OR getanyapi'). You can embed X advanced-search operators directly in the query to filter results: from:username and to:username (author or recipient), since:YYYY-MM-DD and until:YYYY-MM-DD (date range), min_faves:N, min_retweets:N, min_replies:N (engagement floors), "exact phrase", -term to exclude, filter:media and filter:links and -filter:replies (content filters), lang:en, near:city, and geocode:lat,long,radius. Examples: 'from:OpenAI', 'AI agents min_faves:500 -filter:replies', 'nvidia since:2026-01-01 until:2026-03-01'. A query with no matches returns an empty items array; prefer the fewest words that identify the topic.
      */
@@ -39124,6 +40689,11 @@ interface TwitterSearchCommunitiesInput {
      * Opaque pagination cursor from a previous response's nextCursor. Omit for the first page; pass it to fetch the next page of communities.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to match against X (Twitter) community names and topics (e.g. artificial intelligence).
      */
@@ -39178,6 +40748,11 @@ interface TwitterSearchUsersInput {
      * Opaque pagination cursor from a previous response's nextCursor. Omit for the first page; pass it to fetch the next page of accounts.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keyword to match against X (Twitter) handles, display names, and profile bios. This is the People tab of X search: pass a topic ('ai agents'), a name ('Elon Musk'), or a handle ('openai') and get back accounts, not posts. Use twitter.search for posts.
      */
@@ -39252,6 +40827,11 @@ interface TwitterSearchUsersData {
  * Input for X / Twitter Tweet Thread (twitter.thread).
  */
 interface TwitterThreadInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Canonical x.com or twitter.com status URL with a numeric tweet ID. Any tweet in the self-thread can be supplied; the provider resolves the thread root.
      */
@@ -39365,6 +40945,11 @@ interface TwitterTrendsInput {
      * Country name, city name, or ISO country code. Omit for worldwide trends.
      */
     location?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TwitterTrendsItem {
     /**
@@ -39415,6 +41000,11 @@ interface TwitterTrendsData {
  * Input for Twitter Tweet (twitter.tweet).
  */
 interface TwitterTweetInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Canonical x.com or twitter.com status URL with a numeric tweet ID, including /i/web/status and media-share variants.
      */
@@ -39484,6 +41074,11 @@ interface TwitterTweetData {
  * Input for Twitter Tweet Transcript (twitter.tweet_transcript).
  */
 interface TwitterTweetTranscriptInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Tweet URL of the video to transcribe (e.g. https://x.com/TheoVon/status/1916982720317821050).
      */
@@ -39508,6 +41103,11 @@ interface TwitterUserPostsInput {
      * Twitter/X handle without the leading @.
      */
     handle: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface TwitterUserPostsTweet {
     /**
@@ -39619,6 +41219,11 @@ interface TwitterUserTweetsInput {
      * Default: 20.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Require a lane that can return the requested limit in one response.
      */
@@ -39957,6 +41562,11 @@ interface UpworkJobsInput {
      * When true, only return jobs from clients with a verified payment method.
      */
     paymentVerified?: boolean;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Keywords to search Upwork jobs for (e.g. react developer).
      */
@@ -40061,6 +41671,11 @@ declare class UpworkNamespace {
  * Input for Walmart Product (walmart.product).
  */
 interface WalmartProductInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Walmart product page URL.
      */
@@ -40169,6 +41784,11 @@ interface WebCrawlInput {
      * Range: minimum 1, maximum 10.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The website URL or domain to crawl.
      */
@@ -40207,6 +41827,11 @@ interface WebMapInput {
      * Default: 100.
      */
     limit?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Optional term that orders the returned links by relevance.
      */
@@ -40269,6 +41894,11 @@ interface WebScrapeInput {
      * Default: false.
      */
     onlyMainContent?: boolean;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * When true, fetch through a stealth proxy that gets past bot protection which refuses an ordinary request. This costs materially more per request (see the pricing ceiling), so leave it off unless a normal scrape of the site comes back blocked or empty.
      * Default: false.
@@ -40319,6 +41949,11 @@ interface WebScrapeData {
  * Input for Website Screenshot (web.screenshot).
  */
 interface WebScreenshotInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The full URL of the page to capture.
      */
@@ -40407,6 +42042,11 @@ declare class WebNamespace {
  * Input for Weibo Hot Search (weibo.hot_search).
  */
 interface WeiboHotSearchInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface WeiboHotSearchTopic {
     /**
@@ -40457,6 +42097,11 @@ interface WeiboPostInput {
      * Weibo post identifier.
      */
     postId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Weibo Post (weibo.post).
@@ -40539,6 +42184,11 @@ interface WeiboPostCommentsInput {
      * Weibo post identifier.
      */
     postId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface WeiboPostCommentsComment {
     /**
@@ -40601,6 +42251,11 @@ interface WeiboPostCommentsData {
  * Input for Weibo Profile (weibo.profile).
  */
 interface WeiboProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Weibo user identifier.
      */
@@ -40681,6 +42336,11 @@ interface WeiboSearchInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keyword.
      */
@@ -40786,6 +42446,11 @@ interface WeiboUserPostsInput {
      * Default: 1.
      */
     page?: number;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Weibo user identifier.
      */
@@ -40955,6 +42620,11 @@ declare class WeiboNamespace {
  * Input for Yahoo Finance Quote (yahoo_finance.quote).
  */
 interface YahooFinanceQuoteInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The ticker symbol to look up.
      */
@@ -41047,6 +42717,11 @@ interface YelpSearchInput {
      * City and state defining the search area (e.g. San Francisco, CA).
      */
     location: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search term or category to look for (e.g. sushi).
      */
@@ -41200,6 +42875,11 @@ interface YoutubeChannelInput {
      * YouTube channel handle.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of YouTube Channel (youtube.channel).
@@ -41239,6 +42919,11 @@ interface YoutubeChannelCommunityPostsInput {
      * YouTube channel handle.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface YoutubeChannelCommunityPostsPost {
     /**
@@ -41290,6 +42975,11 @@ interface YoutubeChannelLivesInput {
      * YouTube channel handle.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface YoutubeChannelLivesLive {
     /**
@@ -41341,6 +43031,11 @@ interface YoutubeChannelPlaylistsInput {
      * YouTube channel handle.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface YoutubeChannelPlaylistsPlaylist {
     /**
@@ -41388,6 +43083,11 @@ interface YoutubeChannelShortsInput {
      * YouTube channel handle, including or omitting the leading @.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Sort order for the Shorts feed. latest and newest are equivalent.
      * One of: latest, newest, popular.
@@ -41466,6 +43166,11 @@ interface YoutubeChannelVideosInput {
      * YouTube channel handle.
      */
     handle?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Sort order.
      * One of: latest, popular.
@@ -41514,6 +43219,11 @@ interface YoutubeCommentRepliesInput {
      * Replies continuation token from the comments endpoint, or the continuationToken from a previous replies response for further pagination.
      */
     continuationToken: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface YoutubeCommentRepliesComment {
     /**
@@ -41549,6 +43259,11 @@ interface YoutubeCommentRepliesData {
  * Input for YouTube Community Post (youtube.community_post).
  */
 interface YoutubeCommunityPostInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * URL of the YouTube community post.
      */
@@ -41588,6 +43303,11 @@ interface YoutubePlaylistInput {
      * The playlist ID: the "list" parameter in a playlist URL (e.g. "PLu0W_9lII9ahIappRPN0MCAgtOu3lQjQi").
      */
     playlistId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface YoutubePlaylistVideo {
     /**
@@ -41640,6 +43360,11 @@ interface YoutubeSearchInput {
      * Continuation token from a previous response for pagination.
      */
     cursor?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * The YouTube search query.
      */
@@ -41710,6 +43435,11 @@ interface YoutubeSearchHashtagInput {
      * Hashtag to search for (without the leading #).
      */
     hashtag: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Content filter.
      * One of: all, shorts.
@@ -41758,6 +43488,11 @@ interface YoutubeSearchHashtagData {
  * Input for YouTube Trending Shorts (youtube.trending_shorts).
  */
 interface YoutubeTrendingShortsInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 interface YoutubeTrendingShortsShort {
     /**
@@ -41801,6 +43536,11 @@ interface YoutubeVideoInput {
      * YouTube video ID.
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full YouTube video URL.
      */
@@ -41856,6 +43596,11 @@ interface YoutubeVideoCommentsInput {
      * Comment order (e.g. top, newest).
      */
     order?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full YouTube video URL.
      */
@@ -41903,6 +43648,11 @@ interface YoutubeVideoSponsorsInput {
      * 2-letter language code for transcript lookup (e.g. en, es, fr).
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * YouTube video or Short URL.
      */
@@ -41941,6 +43691,11 @@ interface YoutubeVideoTranscriptInput {
      * YouTube video ID.
      */
     id?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Full YouTube video URL.
      */
@@ -41995,6 +43750,11 @@ interface YoutubeVideoTranscriptFullInput {
      * Preferred caption language code (e.g. "en", "es"). Defaults to English.
      */
     language?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * YouTube video URL (e.g. "https://www.youtube.com/watch?v=dQw4w9WgXcQ").
      */
@@ -42317,6 +44077,11 @@ interface ZhihuAnswerInput {
      * Zhihu answer identifier.
      */
     answerId: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
 }
 /**
  * The `data` payload of Zhihu Answer (zhihu.answer).
@@ -42383,6 +44148,11 @@ interface ZhihuAnswerData {
  * Input for Zhihu Profile (zhihu.profile).
  */
 interface ZhihuProfileInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Zhihu user URL token.
      */
@@ -42451,6 +44221,11 @@ interface ZhihuProfileData {
  * Input for Zhihu Question (zhihu.question).
  */
 interface ZhihuQuestionInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Zhihu question identifier.
      */
@@ -42535,6 +44310,11 @@ interface ZhihuQuestionAnswersInput {
      * Default: default.
      */
     order?: "default" | "updated";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Zhihu question identifier.
      */
@@ -42640,6 +44420,11 @@ interface ZhihuSearchArticlesInput {
      * Default: 0.
      */
     offset?: string;
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Search keyword.
      */
@@ -42821,6 +44606,11 @@ declare class ZhihuNamespace {
  * Input for Zillow Property (zillow.property).
  */
 interface ZillowPropertyInput {
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Zillow property details URL (e.g. https://www.zillow.com/homedetails/123-Main-St-Anytown-CA-90210/12345678_zpid/).
      */
@@ -42993,6 +44783,11 @@ interface ZillowSearchInput {
      * Default: buy.
      */
     operation?: "buy" | "rent" | "sold";
+    /**
+     * Optional. Prefer sources whose typical response time (median over the trailing 30 days, as published on this endpoint's lane health) is under this many milliseconds; among those, the cheapest serves. If no source is that fast the request is still served, by whichever source offers the best speed for its price - it is never refused for being slow. Sources we have not timed are tried last. This is a preference, not a guarantee: the median describes past requests and is not a ceiling on this one, and it excludes any wait this request itself asks for. On a paginated walk it applies to the first page only: later pages stay with the source that page chose, at the price it was quoted.
+     * Range: minimum 1.
+     */
+    preferLatencyUnderMs?: number;
     /**
      * Only show listings with a price reduction. Buy searches only; ignored for rentals (e.g. true).
      */