@openrouter/sdk 0.3.11 → 0.3.14

package/esm/models/openresponsesrequest.d.ts

@@ -12,10 +12,13 @@ import { OpenResponsesWebSearchPreview20250311Tool, OpenResponsesWebSearchPrevie
 import { OpenResponsesWebSearchPreviewTool, OpenResponsesWebSearchPreviewTool$Outbound } from "./openresponseswebsearchpreviewtool.js";
 import { OpenResponsesWebSearchTool, OpenResponsesWebSearchTool$Outbound } from "./openresponseswebsearchtool.js";
 import { PDFParserOptions, PDFParserOptions$Outbound } from "./pdfparseroptions.js";
+import { PreferredMaxLatency, PreferredMaxLatency$Outbound } from "./preferredmaxlatency.js";
+import { PreferredMinThroughput, PreferredMinThroughput$Outbound } from "./preferredminthroughput.js";
 import { ProviderName } from "./providername.js";
 import { ProviderSort } from "./providersort.js";
 import { ProviderSortConfig, ProviderSortConfig$Outbound } from "./providersortconfig.js";
 import { Quantization } from "./quantization.js";
+import { ResponsesOutputModality } from "./responsesoutputmodality.js";
 import { WebSearchEngine } from "./websearchengine.js";
 /**
  * Function tool definition
@@ -30,6 +33,7 @@ export type OpenResponsesRequestToolFunction = {
     } | null;
 };
 export type OpenResponsesRequestToolUnion = OpenResponsesRequestToolFunction | OpenResponsesWebSearchPreviewTool | OpenResponsesWebSearchPreview20250311Tool | OpenResponsesWebSearchTool | OpenResponsesWebSearch20250826Tool;
+export type OpenResponsesRequestImageConfig = string | number;
 export declare const ServiceTier: {
     readonly Auto: "auto";
 };
@@ -129,25 +133,13 @@ export type OpenResponsesRequestProvider = {
      */
     maxPrice?: OpenResponsesRequestMaxPrice | undefined;
     /**
-     * Preferred minimum throughput (in tokens per second). Endpoints below this threshold may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
+     * Preferred minimum throughput (in tokens per second). Can be a number (applies to p50) or an object with percentile-specific cutoffs. Endpoints below the threshold(s) may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
      */
-    preferredMinThroughput?: number | null | undefined;
+    preferredMinThroughput?: PreferredMinThroughput | null | undefined;
     /**
-     * Preferred maximum latency (in seconds). Endpoints above this threshold may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
+     * Preferred maximum latency (in seconds). Can be a number (applies to p50) or an object with percentile-specific cutoffs. Endpoints above the threshold(s) may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
      */
-    preferredMaxLatency?: number | null | undefined;
-    /**
-     * **DEPRECATED** Use preferred_min_throughput instead. Backwards-compatible alias for preferred_min_throughput.
-     *
-     * @deprecated field: Use preferred_min_throughput instead..
-     */
-    minThroughput?: number | null | undefined;
-    /**
-     * **DEPRECATED** Use preferred_max_latency instead. Backwards-compatible alias for preferred_max_latency.
-     *
-     * @deprecated field: Use preferred_max_latency instead..
-     */
-    maxLatency?: number | null | undefined;
+    preferredMaxLatency?: PreferredMaxLatency | null | undefined;
 };
 export type OpenResponsesRequestPluginResponseHealing = {
     id: "response-healing";
@@ -183,7 +175,18 @@ export type OpenResponsesRequestPluginWeb = {
 export type OpenResponsesRequestPluginModeration = {
     id: "moderation";
 };
-export type OpenResponsesRequestPluginUnion = OpenResponsesRequestPluginModeration | OpenResponsesRequestPluginWeb | OpenResponsesRequestPluginFileParser | OpenResponsesRequestPluginResponseHealing;
+export type OpenResponsesRequestPluginAutoRouter = {
+    id: "auto-router";
+    /**
+     * Set to false to disable the auto-router plugin for this request. Defaults to true.
+     */
+    enabled?: boolean | undefined;
+    /**
+     * List of model patterns to filter which models the auto-router can route between. Supports wildcards (e.g., "anthropic/*" matches all Anthropic models). When not specified, uses the default supported models list.
+     */
+    allowedModels?: Array<string> | undefined;
+};
+export type OpenResponsesRequestPluginUnion = OpenResponsesRequestPluginAutoRouter | OpenResponsesRequestPluginModeration | OpenResponsesRequestPluginWeb | OpenResponsesRequestPluginFileParser | OpenResponsesRequestPluginResponseHealing;
 /**
  * Request schema for Responses endpoint
  */
@@ -215,7 +218,21 @@ export type OpenResponsesRequest = {
     maxOutputTokens?: number | null | undefined;
     temperature?: number | null | undefined;
     topP?: number | null | undefined;
+    topLogprobs?: number | null | undefined;
+    maxToolCalls?: number | null | undefined;
+    presencePenalty?: number | null | undefined;
+    frequencyPenalty?: number | null | undefined;
     topK?: number | undefined;
+    /**
+     * Provider-specific image configuration options. Keys and values vary by model/provider. See https://openrouter.ai/docs/features/multimodal/image-generation for more details.
+     */
+    imageConfig?: {
+        [k: string]: string | number;
+    } | undefined;
+    /**
+     * Output modalities for the response. Supported values are "text" and "image".
+     */
+    modalities?: Array<ResponsesOutputModality> | undefined;
     promptCacheKey?: string | null | undefined;
     previousResponseId?: string | null | undefined;
     prompt?: OpenAIResponsesPrompt | null | undefined;
@@ -233,7 +250,7 @@ export type OpenResponsesRequest = {
     /**
      * Plugins you want to enable for this request, including their settings.
      */
-    plugins?: Array<OpenResponsesRequestPluginModeration | OpenResponsesRequestPluginWeb | OpenResponsesRequestPluginFileParser | OpenResponsesRequestPluginResponseHealing> | undefined;
+    plugins?: Array<OpenResponsesRequestPluginAutoRouter | OpenResponsesRequestPluginModeration | OpenResponsesRequestPluginWeb | OpenResponsesRequestPluginFileParser | OpenResponsesRequestPluginResponseHealing> | undefined;
     /**
      * A unique identifier representing your end-user, which helps distinguish between different users of your app. This allows your app to identify specific users in case of abuse reports, preventing your entire app from being affected by the actions of individual users. Maximum of 128 characters.
      */
@@ -262,6 +279,11 @@ export type OpenResponsesRequestToolUnion$Outbound = OpenResponsesRequestToolFun
 export declare const OpenResponsesRequestToolUnion$outboundSchema: z.ZodType<OpenResponsesRequestToolUnion$Outbound, OpenResponsesRequestToolUnion>;
 export declare function openResponsesRequestToolUnionToJSON(openResponsesRequestToolUnion: OpenResponsesRequestToolUnion): string;
 /** @internal */
+export type OpenResponsesRequestImageConfig$Outbound = string | number;
+/** @internal */
+export declare const OpenResponsesRequestImageConfig$outboundSchema: z.ZodType<OpenResponsesRequestImageConfig$Outbound, OpenResponsesRequestImageConfig>;
+export declare function openResponsesRequestImageConfigToJSON(openResponsesRequestImageConfig: OpenResponsesRequestImageConfig): string;
+/** @internal */
 export declare const ServiceTier$outboundSchema: z.ZodEnum<typeof ServiceTier>;
 /** @internal */
 export declare const Truncation$outboundSchema: z.ZodType<string, Truncation>;
@@ -309,10 +331,8 @@ export type OpenResponsesRequestProvider$Outbound = {
     quantizations?: Array<string> | null | undefined;
     sort?: string | ProviderSortConfig$Outbound | any | null | undefined;
     max_price?: OpenResponsesRequestMaxPrice$Outbound | undefined;
-    preferred_min_throughput?: number | null | undefined;
-    preferred_max_latency?: number | null | undefined;
-    min_throughput?: number | null | undefined;
-    max_latency?: number | null | undefined;
+    preferred_min_throughput?: PreferredMinThroughput$Outbound | null | undefined;
+    preferred_max_latency?: PreferredMaxLatency$Outbound | null | undefined;
 };
 /** @internal */
 export declare const OpenResponsesRequestProvider$outboundSchema: z.ZodType<OpenResponsesRequestProvider$Outbound, OpenResponsesRequestProvider>;
@@ -353,7 +373,16 @@ export type OpenResponsesRequestPluginModeration$Outbound = {
 export declare const OpenResponsesRequestPluginModeration$outboundSchema: z.ZodType<OpenResponsesRequestPluginModeration$Outbound, OpenResponsesRequestPluginModeration>;
 export declare function openResponsesRequestPluginModerationToJSON(openResponsesRequestPluginModeration: OpenResponsesRequestPluginModeration): string;
 /** @internal */
-export type OpenResponsesRequestPluginUnion$Outbound = OpenResponsesRequestPluginModeration$Outbound | OpenResponsesRequestPluginWeb$Outbound | OpenResponsesRequestPluginFileParser$Outbound | OpenResponsesRequestPluginResponseHealing$Outbound;
+export type OpenResponsesRequestPluginAutoRouter$Outbound = {
+    id: "auto-router";
+    enabled?: boolean | undefined;
+    allowed_models?: Array<string> | undefined;
+};
+/** @internal */
+export declare const OpenResponsesRequestPluginAutoRouter$outboundSchema: z.ZodType<OpenResponsesRequestPluginAutoRouter$Outbound, OpenResponsesRequestPluginAutoRouter>;
+export declare function openResponsesRequestPluginAutoRouterToJSON(openResponsesRequestPluginAutoRouter: OpenResponsesRequestPluginAutoRouter): string;
+/** @internal */
+export type OpenResponsesRequestPluginUnion$Outbound = OpenResponsesRequestPluginAutoRouter$Outbound | OpenResponsesRequestPluginModeration$Outbound | OpenResponsesRequestPluginWeb$Outbound | OpenResponsesRequestPluginFileParser$Outbound | OpenResponsesRequestPluginResponseHealing$Outbound;
 /** @internal */
 export declare const OpenResponsesRequestPluginUnion$outboundSchema: z.ZodType<OpenResponsesRequestPluginUnion$Outbound, OpenResponsesRequestPluginUnion>;
 export declare function openResponsesRequestPluginUnionToJSON(openResponsesRequestPluginUnion: OpenResponsesRequestPluginUnion): string;
@@ -374,7 +403,15 @@ export type OpenResponsesRequest$Outbound = {
     max_output_tokens?: number | null | undefined;
     temperature?: number | null | undefined;
     top_p?: number | null | undefined;
+    top_logprobs?: number | null | undefined;
+    max_tool_calls?: number | null | undefined;
+    presence_penalty?: number | null | undefined;
+    frequency_penalty?: number | null | undefined;
     top_k?: number | undefined;
+    image_config?: {
+        [k: string]: string | number;
+    } | undefined;
+    modalities?: Array<string> | undefined;
     prompt_cache_key?: string | null | undefined;
     previous_response_id?: string | null | undefined;
     prompt?: OpenAIResponsesPrompt$Outbound | null | undefined;
@@ -386,7 +423,7 @@ export type OpenResponsesRequest$Outbound = {
     truncation?: string | null | undefined;
     stream: boolean;
     provider?: OpenResponsesRequestProvider$Outbound | null | undefined;
-    plugins?: Array<OpenResponsesRequestPluginModeration$Outbound | OpenResponsesRequestPluginWeb$Outbound | OpenResponsesRequestPluginFileParser$Outbound | OpenResponsesRequestPluginResponseHealing$Outbound> | undefined;
+    plugins?: Array<OpenResponsesRequestPluginAutoRouter$Outbound | OpenResponsesRequestPluginModeration$Outbound | OpenResponsesRequestPluginWeb$Outbound | OpenResponsesRequestPluginFileParser$Outbound | OpenResponsesRequestPluginResponseHealing$Outbound> | undefined;
     user?: string | undefined;
     session_id?: string | undefined;
 };

package/esm/models/openresponsesrequest.js

@@ -17,10 +17,13 @@ import { OpenResponsesWebSearchPreview20250311Tool$outboundSchema, } from "./ope
 import { OpenResponsesWebSearchPreviewTool$outboundSchema, } from "./openresponseswebsearchpreviewtool.js";
 import { OpenResponsesWebSearchTool$outboundSchema, } from "./openresponseswebsearchtool.js";
 import { PDFParserOptions$outboundSchema, } from "./pdfparseroptions.js";
+import { PreferredMaxLatency$outboundSchema, } from "./preferredmaxlatency.js";
+import { PreferredMinThroughput$outboundSchema, } from "./preferredminthroughput.js";
 import { ProviderName$outboundSchema } from "./providername.js";
 import { ProviderSort$outboundSchema } from "./providersort.js";
 import { ProviderSortConfig$outboundSchema, } from "./providersortconfig.js";
 import { Quantization$outboundSchema } from "./quantization.js";
+import { ResponsesOutputModality$outboundSchema, } from "./responsesoutputmodality.js";
 import { WebSearchEngine$outboundSchema, } from "./websearchengine.js";
 export const ServiceTier = {
     Auto: "auto",
@@ -52,6 +55,11 @@ export function openResponsesRequestToolUnionToJSON(openResponsesRequestToolUnio
     return JSON.stringify(OpenResponsesRequestToolUnion$outboundSchema.parse(openResponsesRequestToolUnion));
 }
 /** @internal */
+export const OpenResponsesRequestImageConfig$outboundSchema = z.union([z.string(), z.number()]);
+export function openResponsesRequestImageConfigToJSON(openResponsesRequestImageConfig) {
+    return JSON.stringify(OpenResponsesRequestImageConfig$outboundSchema.parse(openResponsesRequestImageConfig));
+}
+/** @internal */
 export const ServiceTier$outboundSchema = z.enum(ServiceTier);
 /** @internal */
 export const Truncation$outboundSchema = openEnums.outboundSchema(Truncation);
@@ -110,10 +118,10 @@ export const OpenResponsesRequestProvider$outboundSchema = z.object({
     ])).optional(),
     maxPrice: z.lazy(() => OpenResponsesRequestMaxPrice$outboundSchema)
         .optional(),
-    preferredMinThroughput: z.nullable(z.number()).optional(),
-    preferredMaxLatency: z.nullable(z.number()).optional(),
-    minThroughput: z.nullable(z.number()).optional(),
-    maxLatency: z.nullable(z.number()).optional(),
+    preferredMinThroughput: z.nullable(PreferredMinThroughput$outboundSchema)
+        .optional(),
+    preferredMaxLatency: z.nullable(PreferredMaxLatency$outboundSchema)
+        .optional(),
 }).transform((v) => {
     return remap$(v, {
         allowFallbacks: "allow_fallbacks",
@@ -123,8 +131,6 @@ export const OpenResponsesRequestProvider$outboundSchema = z.object({
         maxPrice: "max_price",
         preferredMinThroughput: "preferred_min_throughput",
         preferredMaxLatency: "preferred_max_latency",
-        minThroughput: "min_throughput",
-        maxLatency: "max_latency",
     });
 });
 export function openResponsesRequestProviderToJSON(openResponsesRequestProvider) {
@@ -171,7 +177,21 @@ export function openResponsesRequestPluginModerationToJSON(openResponsesRequestP
     return JSON.stringify(OpenResponsesRequestPluginModeration$outboundSchema.parse(openResponsesRequestPluginModeration));
 }
 /** @internal */
+export const OpenResponsesRequestPluginAutoRouter$outboundSchema = z.object({
+    id: z.literal("auto-router"),
+    enabled: z.boolean().optional(),
+    allowedModels: z.array(z.string()).optional(),
+}).transform((v) => {
+    return remap$(v, {
+        allowedModels: "allowed_models",
+    });
+});
+export function openResponsesRequestPluginAutoRouterToJSON(openResponsesRequestPluginAutoRouter) {
+    return JSON.stringify(OpenResponsesRequestPluginAutoRouter$outboundSchema.parse(openResponsesRequestPluginAutoRouter));
+}
+/** @internal */
 export const OpenResponsesRequestPluginUnion$outboundSchema = z.union([
+    z.lazy(() => OpenResponsesRequestPluginAutoRouter$outboundSchema),
     z.lazy(() => OpenResponsesRequestPluginModeration$outboundSchema),
     z.lazy(() => OpenResponsesRequestPluginWeb$outboundSchema),
     z.lazy(() => OpenResponsesRequestPluginFileParser$outboundSchema),
@@ -201,7 +221,14 @@ export const OpenResponsesRequest$outboundSchema = z.object({
     maxOutputTokens: z.nullable(z.number()).optional(),
     temperature: z.nullable(z.number()).optional(),
     topP: z.nullable(z.number()).optional(),
+    topLogprobs: z.nullable(z.int()).optional(),
+    maxToolCalls: z.nullable(z.int()).optional(),
+    presencePenalty: z.nullable(z.number()).optional(),
+    frequencyPenalty: z.nullable(z.number()).optional(),
     topK: z.number().optional(),
+    imageConfig: z.record(z.string(), z.union([z.string(), z.number()]))
+        .optional(),
+    modalities: z.array(ResponsesOutputModality$outboundSchema).optional(),
     promptCacheKey: z.nullable(z.string()).optional(),
     previousResponseId: z.nullable(z.string()).optional(),
     prompt: z.nullable(OpenAIResponsesPrompt$outboundSchema).optional(),
@@ -215,6 +242,7 @@ export const OpenResponsesRequest$outboundSchema = z.object({
     stream: z.boolean().default(false),
     provider: z.nullable(z.lazy(() => OpenResponsesRequestProvider$outboundSchema)).optional(),
     plugins: z.array(z.union([
+        z.lazy(() => OpenResponsesRequestPluginAutoRouter$outboundSchema),
         z.lazy(() => OpenResponsesRequestPluginModeration$outboundSchema),
         z.lazy(() => OpenResponsesRequestPluginWeb$outboundSchema),
         z.lazy(() => OpenResponsesRequestPluginFileParser$outboundSchema),
@@ -228,7 +256,12 @@ export const OpenResponsesRequest$outboundSchema = z.object({
         parallelToolCalls: "parallel_tool_calls",
         maxOutputTokens: "max_output_tokens",
         topP: "top_p",
+        topLogprobs: "top_logprobs",
+        maxToolCalls: "max_tool_calls",
+        presencePenalty: "presence_penalty",
+        frequencyPenalty: "frequency_penalty",
         topK: "top_k",
+        imageConfig: "image_config",
         promptCacheKey: "prompt_cache_key",
         previousResponseId: "previous_response_id",
         safetyIdentifier: "safety_identifier",

package/esm/models/operations/getgeneration.d.ts

@@ -148,6 +148,10 @@ export type GetGenerationData = {
      * Type of API used for the generation
      */
     apiType: ApiType | null;
+    /**
+     * Router used for the request (e.g., openrouter/auto)
+     */
+    router: string | null;
 };
 /**
  * Generation response

package/esm/models/operations/getgeneration.js

@@ -57,6 +57,7 @@ export const GetGenerationData$inboundSchema = z.object({
     native_finish_reason: z.nullable(z.string()),
     external_user: z.nullable(z.string()),
     api_type: z.nullable(ApiType$inboundSchema),
+    router: z.nullable(z.string()),
 }).transform((v) => {
     return remap$(v, {
         "upstream_id": "upstreamId",

package/esm/models/percentilelatencycutoffs.d.ts

@@ -0,0 +1,33 @@
+import * as z from "zod/v4";
+/**
+ * Percentile-based latency cutoffs. All specified cutoffs must be met for an endpoint to be preferred.
+ */
+export type PercentileLatencyCutoffs = {
+    /**
+     * Maximum p50 latency (seconds)
+     */
+    p50?: number | null | undefined;
+    /**
+     * Maximum p75 latency (seconds)
+     */
+    p75?: number | null | undefined;
+    /**
+     * Maximum p90 latency (seconds)
+     */
+    p90?: number | null | undefined;
+    /**
+     * Maximum p99 latency (seconds)
+     */
+    p99?: number | null | undefined;
+};
+/** @internal */
+export type PercentileLatencyCutoffs$Outbound = {
+    p50?: number | null | undefined;
+    p75?: number | null | undefined;
+    p90?: number | null | undefined;
+    p99?: number | null | undefined;
+};
+/** @internal */
+export declare const PercentileLatencyCutoffs$outboundSchema: z.ZodType<PercentileLatencyCutoffs$Outbound, PercentileLatencyCutoffs>;
+export declare function percentileLatencyCutoffsToJSON(percentileLatencyCutoffs: PercentileLatencyCutoffs): string;
+//# sourceMappingURL=percentilelatencycutoffs.d.ts.map

package/esm/models/percentilelatencycutoffs.js

@@ -0,0 +1,16 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: 92e053096488
+ */
+import * as z from "zod/v4";
+/** @internal */
+export const PercentileLatencyCutoffs$outboundSchema = z.object({
+    p50: z.nullable(z.number()).optional(),
+    p75: z.nullable(z.number()).optional(),
+    p90: z.nullable(z.number()).optional(),
+    p99: z.nullable(z.number()).optional(),
+});
+export function percentileLatencyCutoffsToJSON(percentileLatencyCutoffs) {
+    return JSON.stringify(PercentileLatencyCutoffs$outboundSchema.parse(percentileLatencyCutoffs));
+}
+//# sourceMappingURL=percentilelatencycutoffs.js.map

package/esm/models/percentilestats.d.ts

@@ -0,0 +1,28 @@
+import * as z from "zod/v4";
+import { Result as SafeParseResult } from "../types/fp.js";
+import { SDKValidationError } from "./errors/sdkvalidationerror.js";
+/**
+ * Latency percentiles in milliseconds over the last 30 minutes. Latency measures time to first token. Only visible when authenticated with an API key or cookie; returns null for unauthenticated requests.
+ */
+export type PercentileStats = {
+    /**
+     * Median (50th percentile)
+     */
+    p50: number;
+    /**
+     * 75th percentile
+     */
+    p75: number;
+    /**
+     * 90th percentile
+     */
+    p90: number;
+    /**
+     * 99th percentile
+     */
+    p99: number;
+};
+/** @internal */
+export declare const PercentileStats$inboundSchema: z.ZodType<PercentileStats, unknown>;
+export declare function percentileStatsFromJSON(jsonString: string): SafeParseResult<PercentileStats, SDKValidationError>;
+//# sourceMappingURL=percentilestats.d.ts.map

package/esm/models/percentilestats.js

@@ -0,0 +1,17 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: 9a6b504d3964
+ */
+import * as z from "zod/v4";
+import { safeParse } from "../lib/schemas.js";
+/** @internal */
+export const PercentileStats$inboundSchema = z.object({
+    p50: z.number(),
+    p75: z.number(),
+    p90: z.number(),
+    p99: z.number(),
+});
+export function percentileStatsFromJSON(jsonString) {
+    return safeParse(jsonString, (x) => PercentileStats$inboundSchema.parse(JSON.parse(x)), `Failed to parse 'PercentileStats' from JSON`);
+}
+//# sourceMappingURL=percentilestats.js.map

package/esm/models/percentilethroughputcutoffs.d.ts

@@ -0,0 +1,33 @@
+import * as z from "zod/v4";
+/**
+ * Percentile-based throughput cutoffs. All specified cutoffs must be met for an endpoint to be preferred.
+ */
+export type PercentileThroughputCutoffs = {
+    /**
+     * Minimum p50 throughput (tokens/sec)
+     */
+    p50?: number | null | undefined;
+    /**
+     * Minimum p75 throughput (tokens/sec)
+     */
+    p75?: number | null | undefined;
+    /**
+     * Minimum p90 throughput (tokens/sec)
+     */
+    p90?: number | null | undefined;
+    /**
+     * Minimum p99 throughput (tokens/sec)
+     */
+    p99?: number | null | undefined;
+};
+/** @internal */
+export type PercentileThroughputCutoffs$Outbound = {
+    p50?: number | null | undefined;
+    p75?: number | null | undefined;
+    p90?: number | null | undefined;
+    p99?: number | null | undefined;
+};
+/** @internal */
+export declare const PercentileThroughputCutoffs$outboundSchema: z.ZodType<PercentileThroughputCutoffs$Outbound, PercentileThroughputCutoffs>;
+export declare function percentileThroughputCutoffsToJSON(percentileThroughputCutoffs: PercentileThroughputCutoffs): string;
+//# sourceMappingURL=percentilethroughputcutoffs.d.ts.map

package/esm/models/percentilethroughputcutoffs.js

@@ -0,0 +1,16 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: 847aba8ff633
+ */
+import * as z from "zod/v4";
+/** @internal */
+export const PercentileThroughputCutoffs$outboundSchema = z.object({
+    p50: z.nullable(z.number()).optional(),
+    p75: z.nullable(z.number()).optional(),
+    p90: z.nullable(z.number()).optional(),
+    p99: z.nullable(z.number()).optional(),
+});
+export function percentileThroughputCutoffsToJSON(percentileThroughputCutoffs) {
+    return JSON.stringify(PercentileThroughputCutoffs$outboundSchema.parse(percentileThroughputCutoffs));
+}
+//# sourceMappingURL=percentilethroughputcutoffs.js.map

package/esm/models/preferredmaxlatency.d.ts

@@ -0,0 +1,12 @@
+import * as z from "zod/v4";
+import { PercentileLatencyCutoffs, PercentileLatencyCutoffs$Outbound } from "./percentilelatencycutoffs.js";
+/**
+ * Preferred maximum latency (in seconds). Can be a number (applies to p50) or an object with percentile-specific cutoffs. Endpoints above the threshold(s) may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
+ */
+export type PreferredMaxLatency = number | PercentileLatencyCutoffs | any;
+/** @internal */
+export type PreferredMaxLatency$Outbound = number | PercentileLatencyCutoffs$Outbound | any;
+/** @internal */
+export declare const PreferredMaxLatency$outboundSchema: z.ZodType<PreferredMaxLatency$Outbound, PreferredMaxLatency>;
+export declare function preferredMaxLatencyToJSON(preferredMaxLatency: PreferredMaxLatency): string;
+//# sourceMappingURL=preferredmaxlatency.d.ts.map

package/esm/models/preferredmaxlatency.js

@@ -0,0 +1,12 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: e03f33269427
+ */
+import * as z from "zod/v4";
+import { PercentileLatencyCutoffs$outboundSchema, } from "./percentilelatencycutoffs.js";
+/** @internal */
+export const PreferredMaxLatency$outboundSchema = z.union([z.number(), PercentileLatencyCutoffs$outboundSchema, z.any()]);
+export function preferredMaxLatencyToJSON(preferredMaxLatency) {
+    return JSON.stringify(PreferredMaxLatency$outboundSchema.parse(preferredMaxLatency));
+}
+//# sourceMappingURL=preferredmaxlatency.js.map

package/esm/models/preferredminthroughput.d.ts

@@ -0,0 +1,12 @@
+import * as z from "zod/v4";
+import { PercentileThroughputCutoffs, PercentileThroughputCutoffs$Outbound } from "./percentilethroughputcutoffs.js";
+/**
+ * Preferred minimum throughput (in tokens per second). Can be a number (applies to p50) or an object with percentile-specific cutoffs. Endpoints below the threshold(s) may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
+ */
+export type PreferredMinThroughput = number | PercentileThroughputCutoffs | any;
+/** @internal */
+export type PreferredMinThroughput$Outbound = number | PercentileThroughputCutoffs$Outbound | any;
+/** @internal */
+export declare const PreferredMinThroughput$outboundSchema: z.ZodType<PreferredMinThroughput$Outbound, PreferredMinThroughput>;
+export declare function preferredMinThroughputToJSON(preferredMinThroughput: PreferredMinThroughput): string;
+//# sourceMappingURL=preferredminthroughput.d.ts.map

package/esm/models/preferredminthroughput.js

@@ -0,0 +1,12 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: 5ff9056f1474
+ */
+import * as z from "zod/v4";
+import { PercentileThroughputCutoffs$outboundSchema, } from "./percentilethroughputcutoffs.js";
+/** @internal */
+export const PreferredMinThroughput$outboundSchema = z.union([z.number(), PercentileThroughputCutoffs$outboundSchema, z.any()]);
+export function preferredMinThroughputToJSON(preferredMinThroughput) {
+    return JSON.stringify(PreferredMinThroughput$outboundSchema.parse(preferredMinThroughput));
+}
+//# sourceMappingURL=preferredminthroughput.js.map

package/esm/models/providername.d.ts

@@ -27,12 +27,12 @@ export declare const ProviderName: {
     readonly Fireworks: "Fireworks";
     readonly Friendli: "Friendli";
     readonly GMICloud: "GMICloud";
-    readonly GoPomelo: "GoPomelo";
     readonly Google: "Google";
     readonly GoogleAIStudio: "Google AI Studio";
     readonly Groq: "Groq";
     readonly Hyperbolic: "Hyperbolic";
     readonly Inception: "Inception";
+    readonly Inceptron: "Inceptron";
     readonly InferenceNet: "InferenceNet";
     readonly Infermatic: "Infermatic";
     readonly Inflection: "Inflection";
@@ -57,13 +57,14 @@ export declare const ProviderName: {
     readonly Phala: "Phala";
     readonly Relace: "Relace";
     readonly SambaNova: "SambaNova";
+    readonly Seed: "Seed";
     readonly SiliconFlow: "SiliconFlow";
     readonly Sourceful: "Sourceful";
     readonly Stealth: "Stealth";
     readonly StreamLake: "StreamLake";
     readonly Switchpoint: "Switchpoint";
-    readonly Targon: "Targon";
     readonly Together: "Together";
+    readonly Upstage: "Upstage";
     readonly Venice: "Venice";
     readonly WandB: "WandB";
     readonly Xiaomi: "Xiaomi";

package/esm/models/providername.js

@@ -30,12 +30,12 @@ export const ProviderName = {
     Fireworks: "Fireworks",
     Friendli: "Friendli",
     GMICloud: "GMICloud",
-    GoPomelo: "GoPomelo",
     Google: "Google",
     GoogleAIStudio: "Google AI Studio",
     Groq: "Groq",
     Hyperbolic: "Hyperbolic",
     Inception: "Inception",
+    Inceptron: "Inceptron",
     InferenceNet: "InferenceNet",
     Infermatic: "Infermatic",
     Inflection: "Inflection",
@@ -60,13 +60,14 @@ export const ProviderName = {
     Phala: "Phala",
     Relace: "Relace",
     SambaNova: "SambaNova",
+    Seed: "Seed",
     SiliconFlow: "SiliconFlow",
     Sourceful: "Sourceful",
     Stealth: "Stealth",
     StreamLake: "StreamLake",
     Switchpoint: "Switchpoint",
-    Targon: "Targon",
     Together: "Together",
+    Upstage: "Upstage",
     Venice: "Venice",
     WandB: "WandB",
     Xiaomi: "Xiaomi",

package/esm/models/providerpreferences.d.ts

@@ -1,6 +1,8 @@
 import * as z from "zod/v4";
 import { ClosedEnum, OpenEnum } from "../types/enums.js";
 import { DataCollection } from "./datacollection.js";
+import { PreferredMaxLatency, PreferredMaxLatency$Outbound } from "./preferredmaxlatency.js";
+import { PreferredMinThroughput, PreferredMinThroughput$Outbound } from "./preferredminthroughput.js";
 import { ProviderName } from "./providername.js";
 import { ProviderSort } from "./providersort.js";
 import { Quantization } from "./quantization.js";
@@ -119,25 +121,13 @@ export type ProviderPreferences = {
      */
     maxPrice?: ProviderPreferencesMaxPrice | undefined;
     /**
-     * Preferred minimum throughput (in tokens per second). Endpoints below this threshold may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
+     * Preferred minimum throughput (in tokens per second). Can be a number (applies to p50) or an object with percentile-specific cutoffs. Endpoints below the threshold(s) may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
      */
-    preferredMinThroughput?: number | null | undefined;
+    preferredMinThroughput?: PreferredMinThroughput | null | undefined;
     /**
-     * Preferred maximum latency (in seconds). Endpoints above this threshold may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
+     * Preferred maximum latency (in seconds). Can be a number (applies to p50) or an object with percentile-specific cutoffs. Endpoints above the threshold(s) may still be used, but are deprioritized in routing. When using fallback models, this may cause a fallback model to be used instead of the primary model if it meets the threshold.
      */
-    preferredMaxLatency?: number | null | undefined;
-    /**
-     * **DEPRECATED** Use preferred_min_throughput instead. Backwards-compatible alias for preferred_min_throughput.
-     *
-     * @deprecated field: Use preferred_min_throughput instead..
-     */
-    minThroughput?: number | null | undefined;
-    /**
-     * **DEPRECATED** Use preferred_max_latency instead. Backwards-compatible alias for preferred_max_latency.
-     *
-     * @deprecated field: Use preferred_max_latency instead..
-     */
-    maxLatency?: number | null | undefined;
+    preferredMaxLatency?: PreferredMaxLatency | null | undefined;
 };
 /** @internal */
 export type ProviderPreferencesOrder$Outbound = string | string;
@@ -204,10 +194,8 @@ export type ProviderPreferences$Outbound = {
     quantizations?: Array<string> | null | undefined;
     sort?: string | ProviderPreferencesProviderSortConfig$Outbound | string | string | null | undefined;
     max_price?: ProviderPreferencesMaxPrice$Outbound | undefined;
-    preferred_min_throughput?: number | null | undefined;
-    preferred_max_latency?: number | null | undefined;
-    min_throughput?: number | null | undefined;
-    max_latency?: number | null | undefined;
+    preferred_min_throughput?: PreferredMinThroughput$Outbound | null | undefined;
+    preferred_max_latency?: PreferredMaxLatency$Outbound | null | undefined;
 };
 /** @internal */
 export declare const ProviderPreferences$outboundSchema: z.ZodType<ProviderPreferences$Outbound, ProviderPreferences>;

package/esm/models/providerpreferences.js

@@ -6,6 +6,8 @@ import * as z from "zod/v4";
 import { remap as remap$ } from "../lib/primitives.js";
 import * as openEnums from "../types/enums.js";
 import { DataCollection$outboundSchema, } from "./datacollection.js";
+import { PreferredMaxLatency$outboundSchema, } from "./preferredmaxlatency.js";
+import { PreferredMinThroughput$outboundSchema, } from "./preferredminthroughput.js";
 import { ProviderName$outboundSchema } from "./providername.js";
 import { ProviderSort$outboundSchema } from "./providersort.js";
 import { Quantization$outboundSchema } from "./quantization.js";
@@ -113,10 +115,10 @@ export const ProviderPreferences$outboundSchema = z.object({
         SortEnum$outboundSchema,
     ])).optional(),
     maxPrice: z.lazy(() => ProviderPreferencesMaxPrice$outboundSchema).optional(),
-    preferredMinThroughput: z.nullable(z.number()).optional(),
-    preferredMaxLatency: z.nullable(z.number()).optional(),
-    minThroughput: z.nullable(z.number()).optional(),
-    maxLatency: z.nullable(z.number()).optional(),
+    preferredMinThroughput: z.nullable(PreferredMinThroughput$outboundSchema)
+        .optional(),
+    preferredMaxLatency: z.nullable(PreferredMaxLatency$outboundSchema)
+        .optional(),
 }).transform((v) => {
     return remap$(v, {
         allowFallbacks: "allow_fallbacks",
@@ -126,8 +128,6 @@ export const ProviderPreferences$outboundSchema = z.object({
         maxPrice: "max_price",
         preferredMinThroughput: "preferred_min_throughput",
         preferredMaxLatency: "preferred_max_latency",
-        minThroughput: "min_throughput",
-        maxLatency: "max_latency",
     });
 });
 export function providerPreferencesToJSON(providerPreferences) {