@openrouter/sdk 1.4.24 → 1.5.0

package/esm/models/outputwebsearchservertoolitem.d.ts

@@ -2,7 +2,7 @@ import * as z from "zod/v4";
 import { ClosedEnum } from "../types/enums.js";
 import { Result as SafeParseResult } from "../types/fp.js";
 import { SDKValidationError } from "./errors/sdkvalidationerror.js";
-import { ToolCallStatus } from "./toolcallstatus.js";
+import { FailableToolCallStatus } from "./failabletoolcallstatus.js";
 export declare const OutputWebSearchServerToolItemTypeURL: {
     readonly Url: "url";
 };
@@ -35,8 +35,12 @@ export type OutputWebSearchServerToolItem = {
      * The search action performed, matching OpenAI web_search_call.action shape. Includes the query the model issued and optional source URLs returned by the search provider.
      */
     action?: OutputWebSearchServerToolItemAction | undefined;
+    /**
+     * The error message when the tool call failed before producing a result. Set together with `status: 'failed'`; absent on a successful call.
+     */
+    error?: string | undefined;
     id?: string | undefined;
-    status: ToolCallStatus;
+    status: FailableToolCallStatus;
     type: OutputWebSearchServerToolItemTypeOpenrouterWebSearch;
 };
 /** @internal */
@@ -79,6 +83,7 @@ export declare const OutputWebSearchServerToolItem$inboundSchema: z.ZodType<Outp
 /** @internal */
 export type OutputWebSearchServerToolItem$Outbound = {
     action?: OutputWebSearchServerToolItemAction$Outbound | undefined;
+    error?: string | undefined;
     id?: string | undefined;
     status: string;
     type: string;

package/esm/models/outputwebsearchservertoolitem.js

@@ -4,7 +4,7 @@
  */
 import * as z from "zod/v4";
 import { safeParse } from "../lib/schemas.js";
-import { ToolCallStatus$inboundSchema, ToolCallStatus$outboundSchema, } from "./toolcallstatus.js";
+import { FailableToolCallStatus$inboundSchema, FailableToolCallStatus$outboundSchema, } from "./failabletoolcallstatus.js";
 export const OutputWebSearchServerToolItemTypeURL = {
     Url: "url",
 };
@@ -65,16 +65,18 @@ export const OutputWebSearchServerToolItemTypeOpenrouterWebSearch$outboundSchema
 export const OutputWebSearchServerToolItem$inboundSchema = z.object({
     action: z.lazy(() => OutputWebSearchServerToolItemAction$inboundSchema)
         .optional(),
+    error: z.string().optional(),
     id: z.string().optional(),
-    status: ToolCallStatus$inboundSchema,
+    status: FailableToolCallStatus$inboundSchema,
     type: OutputWebSearchServerToolItemTypeOpenrouterWebSearch$inboundSchema,
 });
 /** @internal */
 export const OutputWebSearchServerToolItem$outboundSchema = z.object({
     action: z.lazy(() => OutputWebSearchServerToolItemAction$outboundSchema)
         .optional(),
+    error: z.string().optional(),
     id: z.string().optional(),
-    status: ToolCallStatus$outboundSchema,
+    status: FailableToolCallStatus$outboundSchema,
     type: OutputWebSearchServerToolItemTypeOpenrouterWebSearch$outboundSchema,
 });
 export function outputWebSearchServerToolItemToJSON(outputWebSearchServerToolItem) {

package/esm/models/preview20250311websearchservertool.d.ts

@@ -17,7 +17,7 @@ export type Preview20250311WebSearchServerToolType = ClosedEnum<typeof Preview20
  */
 export type Preview20250311WebSearchServerTool = {
     /**
-     * Which search engine to use. "auto" (default) uses native if the provider supports it, otherwise Exa. "native" forces the provider's built-in search. "exa" forces the Exa search API. "firecrawl" uses Firecrawl (requires BYOK). "parallel" uses the Parallel search API. "perplexity" uses the Perplexity Search API (raw ranked results).
+     * Which search engine to use. "auto" (default) uses native if the provider supports it, otherwise Exa. "native" forces the provider's built-in search. "exa" forces the Exa search API. "firecrawl" uses the Firecrawl search API. "parallel" uses the Parallel search API. "perplexity" uses the Perplexity Search API (raw ranked results).
      */
     engine?: WebSearchEngineEnum | undefined;
     filters?: WebSearchDomainFilter | null | undefined;

package/esm/models/previewwebsearchservertool.d.ts

@@ -17,7 +17,7 @@ export type PreviewWebSearchServerToolType = ClosedEnum<typeof PreviewWebSearchS
  */
 export type PreviewWebSearchServerTool = {
     /**
-     * Which search engine to use. "auto" (default) uses native if the provider supports it, otherwise Exa. "native" forces the provider's built-in search. "exa" forces the Exa search API. "firecrawl" uses Firecrawl (requires BYOK). "parallel" uses the Parallel search API. "perplexity" uses the Perplexity Search API (raw ranked results).
+     * Which search engine to use. "auto" (default) uses native if the provider supports it, otherwise Exa. "native" forces the provider's built-in search. "exa" forces the Exa search API. "firecrawl" uses the Firecrawl search API. "parallel" uses the Parallel search API. "perplexity" uses the Perplexity Search API (raw ranked results).
      */
     engine?: WebSearchEngineEnum | undefined;
     filters?: WebSearchDomainFilter | null | undefined;

package/esm/models/privateendpointpricing.d.ts

@@ -5,6 +5,14 @@ import { SDKValidationError } from "./errors/sdkvalidationerror.js";
  * Negotiated per-token rates reported for requests routed to this endpoint.
  */
 export type PrivateEndpointPricing = {
+    /**
+     * USD per cache-read prompt token, as a decimal string. Omit it to bill cache reads at the provider pricing strategy's default, which for most strategies is the prompt rate.
+     */
+    cacheRead?: string | undefined;
+    /**
+     * USD per cache-write prompt token, as a decimal string. Omit it to bill cache writes at the provider pricing strategy's default.
+     */
+    cacheWrite?: string | undefined;
     /**
      * USD per completion token, as a decimal string.
      */
@@ -18,6 +26,8 @@ export type PrivateEndpointPricing = {
 export declare const PrivateEndpointPricing$inboundSchema: z.ZodType<PrivateEndpointPricing, unknown>;
 /** @internal */
 export type PrivateEndpointPricing$Outbound = {
+    cache_read?: string | undefined;
+    cache_write?: string | undefined;
     completion: string;
     prompt: string;
 };

package/esm/models/privateendpointpricing.js

@@ -3,16 +3,31 @@
  * @generated-id: c780fa3f1dc9
  */
 import * as z from "zod/v4";
+import { remap as remap$ } from "../lib/primitives.js";
 import { safeParse } from "../lib/schemas.js";
 /** @internal */
 export const PrivateEndpointPricing$inboundSchema = z.object({
+    cache_read: z.string().optional(),
+    cache_write: z.string().optional(),
     completion: z.string(),
     prompt: z.string(),
+}).transform((v) => {
+    return remap$(v, {
+        "cache_read": "cacheRead",
+        "cache_write": "cacheWrite",
+    });
 });
 /** @internal */
 export const PrivateEndpointPricing$outboundSchema = z.object({
+    cacheRead: z.string().optional(),
+    cacheWrite: z.string().optional(),
     completion: z.string(),
     prompt: z.string(),
+}).transform((v) => {
+    return remap$(v, {
+        cacheRead: "cache_read",
+        cacheWrite: "cache_write",
+    });
 });
 export function privateEndpointPricingToJSON(privateEndpointPricing) {
     return JSON.stringify(PrivateEndpointPricing$outboundSchema.parse(privateEndpointPricing));

package/esm/models/promptcachebreakpoint.d.ts

@@ -7,10 +7,11 @@ export declare const PromptCacheBreakpointMode: {
 };
 export type PromptCacheBreakpointMode = ClosedEnum<typeof PromptCacheBreakpointMode>;
 /**
- * Marks an explicit prompt-cache boundary on this content block (OpenAI-style). Everything through the block carrying this marker is part of the candidate cached prefix. Supported natively by OpenAI GPT-5.6 and newer; on providers that use Anthropic-style `cache_control`, OpenRouter converts the marker to that format automatically.
+ * Marks an explicit prompt-cache boundary on this content block (OpenAI-style). Everything through the block carrying this marker is part of the candidate cached prefix. Supported natively by OpenAI GPT-5.6 and newer; on providers that use Anthropic-style `cache_control`, OpenRouter converts the marker to that format automatically. `ttl` ("30m" or "12h") overrides the request-level `prompt_cache_options.ttl` for this boundary; 12h writes bill at the higher cache-write rate.
  */
 export type PromptCacheBreakpoint = {
     mode: PromptCacheBreakpointMode;
+    ttl?: string | null | undefined;
 };
 /** @internal */
 export declare const PromptCacheBreakpointMode$inboundSchema: z.ZodEnum<typeof PromptCacheBreakpointMode>;
@@ -21,6 +22,7 @@ export declare const PromptCacheBreakpoint$inboundSchema: z.ZodType<PromptCacheB
 /** @internal */
 export type PromptCacheBreakpoint$Outbound = {
     mode: string;
+    ttl?: string | null | undefined;
 };
 /** @internal */
 export declare const PromptCacheBreakpoint$outboundSchema: z.ZodType<PromptCacheBreakpoint$Outbound, PromptCacheBreakpoint>;

package/esm/models/promptcachebreakpoint.js

@@ -14,10 +14,12 @@ export const PromptCacheBreakpointMode$outboundSchema = PromptCacheBreakpointMod
 /** @internal */
 export const PromptCacheBreakpoint$inboundSchema = z.object({
     mode: PromptCacheBreakpointMode$inboundSchema,
+    ttl: z.nullable(z.string()).optional(),
 });
 /** @internal */
 export const PromptCacheBreakpoint$outboundSchema = z.object({
     mode: PromptCacheBreakpointMode$outboundSchema,
+    ttl: z.nullable(z.string()).optional(),
 });
 export function promptCacheBreakpointToJSON(promptCacheBreakpoint) {
     return JSON.stringify(PromptCacheBreakpoint$outboundSchema.parse(promptCacheBreakpoint));

package/esm/models/promptcacheoptions.d.ts

@@ -7,7 +7,7 @@ export declare const PromptCacheOptionsMode: {
 };
 export type PromptCacheOptionsMode = ClosedEnum<typeof PromptCacheOptionsMode>;
 /**
- * Request-level prompt-cache controls. `mode: "explicit"` disables OpenAI-managed breakpoints so only blocks marked with `prompt_cache_breakpoint` are cached. Only supported by OpenAI GPT-5.6 and newer.
+ * Request-level prompt-cache controls. `mode: "explicit"` disables OpenAI-managed breakpoints so only blocks marked with `prompt_cache_breakpoint` are cached. `ttl` sets the minimum cache lifetime for every breakpoint the request writes: "30m" (default) or "12h" (billed at the higher 12h cache-write rate). Only supported by OpenAI GPT-5.6 and newer.
  */
 export type PromptCacheOptions = {
     mode: PromptCacheOptionsMode;

package/esm/models/providername.d.ts

@@ -79,12 +79,14 @@ export declare const ProviderName: {
     readonly OpenAI: "OpenAI";
     readonly OpenInference: "OpenInference";
     readonly Parasail: "Parasail";
+    readonly ParetoAlpha: "ParetoAlpha";
     readonly Poolside: "Poolside";
     readonly PrimeIntellect: "PrimeIntellect";
     readonly Perceptron: "Perceptron";
     readonly Perplexity: "Perplexity";
     readonly Phala: "Phala";
     readonly Recraft: "Recraft";
+    readonly Recursive: "Recursive";
     readonly Reka: "Reka";
     readonly Relace: "Relace";
     readonly Respan: "Respan";

package/esm/models/providername.js

@@ -82,12 +82,14 @@ export const ProviderName = {
     OpenAI: "OpenAI",
     OpenInference: "OpenInference",
     Parasail: "Parasail",
+    ParetoAlpha: "ParetoAlpha",
     Poolside: "Poolside",
     PrimeIntellect: "PrimeIntellect",
     Perceptron: "Perceptron",
     Perplexity: "Perplexity",
     Phala: "Phala",
     Recraft: "Recraft",
+    Recursive: "Recursive",
     Reka: "Reka",
     Relace: "Relace",
     Respan: "Respan",

package/esm/models/provideroptions.d.ts

@@ -303,6 +303,9 @@ export type ProviderOptions = {
     parasail?: {
         [k: string]: any;
     } | undefined;
+    paretoAlpha?: {
+        [k: string]: any;
+    } | undefined;
     perceptron?: {
         [k: string]: any;
     } | undefined;
@@ -327,6 +330,9 @@ export type ProviderOptions = {
     recursal?: {
         [k: string]: any;
     } | undefined;
+    recursive?: {
+        [k: string]: any;
+    } | undefined;
     reflection?: {
         [k: string]: any;
     } | undefined;
@@ -744,6 +750,9 @@ export type ProviderOptions$Outbound = {
     parasail?: {
         [k: string]: any;
     } | undefined;
+    "pareto-alpha"?: {
+        [k: string]: any;
+    } | undefined;
     perceptron?: {
         [k: string]: any;
     } | undefined;
@@ -768,6 +777,9 @@ export type ProviderOptions$Outbound = {
     recursal?: {
         [k: string]: any;
     } | undefined;
+    recursive?: {
+        [k: string]: any;
+    } | undefined;
     reflection?: {
         [k: string]: any;
     } | undefined;

package/esm/models/provideroptions.js

@@ -106,6 +106,7 @@ export const ProviderOptions$outboundSchema = z.object({
     openInference: z.record(z.string(), z.any()).optional(),
     openai: z.record(z.string(), z.any()).optional(),
     parasail: z.record(z.string(), z.any()).optional(),
+    paretoAlpha: z.record(z.string(), z.any()).optional(),
     perceptron: z.record(z.string(), z.any()).optional(),
     perplexity: z.record(z.string(), z.any()).optional(),
     phala: z.record(z.string(), z.any()).optional(),
@@ -114,6 +115,7 @@ export const ProviderOptions$outboundSchema = z.object({
     quiver: z.record(z.string(), z.any()).optional(),
     recraft: z.record(z.string(), z.any()).optional(),
     recursal: z.record(z.string(), z.any()).optional(),
+    recursive: z.record(z.string(), z.any()).optional(),
     reflection: z.record(z.string(), z.any()).optional(),
     reka: z.record(z.string(), z.any()).optional(),
     relace: z.record(z.string(), z.any()).optional(),
@@ -177,6 +179,7 @@ export const ProviderOptions$outboundSchema = z.object({
         nearAi: "near-ai",
         nexAgi: "nex-agi",
         openInference: "open-inference",
+        paretoAlpha: "pareto-alpha",
         sailResearch: "sail-research",
         sakanaAi: "sakana-ai",
         sambanovaCloaked: "sambanova-cloaked",

package/esm/models/providerresponse.d.ts

@@ -112,12 +112,14 @@ export declare const ProviderResponseProviderName: {
     readonly OpenAI: "OpenAI";
     readonly OpenInference: "OpenInference";
     readonly Parasail: "Parasail";
+    readonly ParetoAlpha: "ParetoAlpha";
     readonly Poolside: "Poolside";
     readonly PrimeIntellect: "PrimeIntellect";
     readonly Perceptron: "Perceptron";
     readonly Perplexity: "Perplexity";
     readonly Phala: "Phala";
     readonly Recraft: "Recraft";
+    readonly Recursive: "Recursive";
     readonly Reka: "Reka";
     readonly Relace: "Relace";
     readonly Respan: "Respan";

package/esm/models/providerresponse.js

@@ -116,12 +116,14 @@ export const ProviderResponseProviderName = {
     OpenAI: "OpenAI",
     OpenInference: "OpenInference",
     Parasail: "Parasail",
+    ParetoAlpha: "ParetoAlpha",
     Poolside: "Poolside",
     PrimeIntellect: "PrimeIntellect",
     Perceptron: "Perceptron",
     Perplexity: "Perplexity",
     Phala: "Phala",
     Recraft: "Recraft",
+    Recursive: "Recursive",
     Reka: "Reka",
     Relace: "Relace",
     Respan: "Respan",

package/esm/models/publicendpoint.d.ts

@@ -6,6 +6,7 @@ import { Parameter } from "./parameter.js";
 import { PercentileStats } from "./percentilestats.js";
 import { PricingOverride } from "./pricingoverride.js";
 import { ProviderName } from "./providername.js";
+import { PublicServiceTier } from "./publicservicetier.js";
 import { Quantization } from "./quantization.js";
 import { ToolChoiceSupport } from "./toolchoicesupport.js";
 export type NativeTools = {
@@ -199,8 +200,16 @@ export type PublicEndpoint = {
     pricing: Pricing;
     providerName: ProviderName;
     quantization: Quantization | null;
+    /**
+     * The service tier this endpoint sells, or null for a standard endpoint. The same value whether the provider selects the tier with a request parameter or with a dedicated model id. Always the canonical tier: priority covers the fast request alias and the turbo, highspeed, and prime provider model-id spellings, none of which this field reports.
+     */
+    serviceTier: PublicServiceTier | null;
     status?: EndpointStatus | undefined;
     supportedParameters: Array<Parameter>;
+    /**
+     * Whether this STT endpoint can label each transcript segment with a speaker (speaker diarization). False when the endpoint has not declared support.
+     */
+    supportsDiarization: boolean;
     /**
      * Whether this TTS endpoint accepts an `image_url` reference describing the desired voice. Requests carrying an image reference are only routed to endpoints where this is true.
      */
@@ -221,12 +230,12 @@ export type PublicEndpoint = {
     tag: string;
     throughputLast30m: PercentileStats | null;
     /**
-     * Uptime percentage over the last 1 day, calculated as successful requests / (successful + error requests) * 100. Rate-limited requests are excluded. Returns null if insufficient data.
+     * Uptime percentage over the last day: the share of minutes in which at least 80% of provider attempts succeeded, counting only minutes with 10 or more attempts. Rate-limited and caller-caused failures are excluded. Null when no minute had enough traffic.
      */
     uptimeLast1d: number | null;
     uptimeLast30m: number | null;
     /**
-     * Uptime percentage over the last 5 minutes, calculated as successful requests / (successful + error requests) * 100. Rate-limited requests are excluded. Returns null if insufficient data.
+     * Uptime percentage over the last 5 minutes: the share of minutes in which at least 80% of provider attempts succeeded, counting only minutes with 10 or more attempts. Rate-limited and caller-caused failures are excluded. Null when no minute had enough traffic.
      */
     uptimeLast5m: number | null;
 };

package/esm/models/publicendpoint.js

@@ -10,6 +10,7 @@ import { Parameter$inboundSchema } from "./parameter.js";
 import { PercentileStats$inboundSchema, } from "./percentilestats.js";
 import { PricingOverride$inboundSchema, } from "./pricingoverride.js";
 import { ProviderName$inboundSchema } from "./providername.js";
+import { PublicServiceTier$inboundSchema, } from "./publicservicetier.js";
 import { Quantization$inboundSchema } from "./quantization.js";
 import { ToolChoiceSupport$inboundSchema, } from "./toolchoicesupport.js";
 /** @internal */
@@ -208,8 +209,10 @@ export const PublicEndpoint$inboundSchema = z.object({
     pricing: z.lazy(() => Pricing$inboundSchema),
     provider_name: ProviderName$inboundSchema,
     quantization: z.nullable(Quantization$inboundSchema),
+    service_tier: z.nullable(PublicServiceTier$inboundSchema),
     status: EndpointStatus$inboundSchema.optional(),
     supported_parameters: z.array(Parameter$inboundSchema),
+    supports_diarization: z.boolean(),
     supports_image_reference: z.boolean().default(false),
     supports_implicit_caching: z.boolean(),
     supports_multiple_audio_references: z.boolean().default(false),
@@ -231,7 +234,9 @@ export const PublicEndpoint$inboundSchema = z.object({
         "native_tools": "nativeTools",
         "perf_last_30m_by_workload": "perfLast30mByWorkload",
         "provider_name": "providerName",
+        "service_tier": "serviceTier",
         "supported_parameters": "supportedParameters",
+        "supports_diarization": "supportsDiarization",
         "supports_image_reference": "supportsImageReference",
         "supports_implicit_caching": "supportsImplicitCaching",
         "supports_multiple_audio_references": "supportsMultipleAudioReferences",

package/esm/models/publicservicetier.d.ts

@@ -0,0 +1,17 @@
+import * as z from "zod/v4";
+import { OpenEnum } from "../types/enums.js";
+/**
+ * The service tier this endpoint sells, or null for a standard endpoint. The same value whether the provider selects the tier with a request parameter or with a dedicated model id. Always the canonical tier: priority covers the fast request alias and the turbo, highspeed, and prime provider model-id spellings, none of which this field reports.
+ */
+export declare const PublicServiceTier: {
+    readonly Flex: "flex";
+    readonly Priority: "priority";
+    readonly Ultrafast: "ultrafast";
+};
+/**
+ * The service tier this endpoint sells, or null for a standard endpoint. The same value whether the provider selects the tier with a request parameter or with a dedicated model id. Always the canonical tier: priority covers the fast request alias and the turbo, highspeed, and prime provider model-id spellings, none of which this field reports.
+ */
+export type PublicServiceTier = OpenEnum<typeof PublicServiceTier>;
+/** @internal */
+export declare const PublicServiceTier$inboundSchema: z.ZodType<PublicServiceTier, unknown>;
+//# sourceMappingURL=publicservicetier.d.ts.map

package/esm/models/publicservicetier.js

@@ -0,0 +1,16 @@
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ * @generated-id: b2568abb5c38
+ */
+import * as openEnums from "../types/enums.js";
+/**
+ * The service tier this endpoint sells, or null for a standard endpoint. The same value whether the provider selects the tier with a request parameter or with a dedicated model id. Always the canonical tier: priority covers the fast request alias and the turbo, highspeed, and prime provider model-id spellings, none of which this field reports.
+ */
+export const PublicServiceTier = {
+    Flex: "flex",
+    Priority: "priority",
+    Ultrafast: "ultrafast",
+};
+/** @internal */
+export const PublicServiceTier$inboundSchema = openEnums.inboundSchema(PublicServiceTier);
+//# sourceMappingURL=publicservicetier.js.map

package/esm/models/responsesrequest.d.ts

@@ -5,6 +5,7 @@ import { AlignmentPlugin, AlignmentPlugin$Outbound } from "./alignmentplugin.js"
 import { AnthropicCacheControlDirective, AnthropicCacheControlDirective$Outbound } from "./anthropiccachecontroldirective.js";
 import { ApplyPatchServerTool, ApplyPatchServerTool$Outbound } from "./applypatchservertool.js";
 import { ApplyPatchServerToolOpenRouter, ApplyPatchServerToolOpenRouter$Outbound } from "./applypatchservertoolopenrouter.js";
+import { ArmorPlugin, ArmorPlugin$Outbound } from "./armorplugin.js";
 import { AutoBetaRouterPlugin, AutoBetaRouterPlugin$Outbound } from "./autobetarouterplugin.js";
 import { AutoRouterPlugin, AutoRouterPlugin$Outbound } from "./autorouterplugin.js";
 import { BashServerTool, BashServerTool$Outbound } from "./bashservertool.js";
@@ -15,6 +16,7 @@ import { ComputerUseServerTool, ComputerUseServerTool$Outbound } from "./compute
 import { ContextCompressionPlugin, ContextCompressionPlugin$Outbound } from "./contextcompressionplugin.js";
 import { CustomTool, CustomTool$Outbound } from "./customtool.js";
 import { DatetimeServerTool, DatetimeServerTool$Outbound } from "./datetimeservertool.js";
+import { DeferredToolsControl, DeferredToolsControl$Outbound } from "./deferredtoolscontrol.js";
 import { FileParserPlugin, FileParserPlugin$Outbound } from "./fileparserplugin.js";
 import { FileSearchServerTool, FileSearchServerTool$Outbound } from "./filesearchservertool.js";
 import { FilesServerTool, FilesServerTool$Outbound } from "./filesservertool.js";
@@ -58,7 +60,7 @@ import { WebFetchServerTool, WebFetchServerTool$Outbound } from "./webfetchserve
 import { WebSearchPlugin, WebSearchPlugin$Outbound } from "./websearchplugin.js";
 import { WebSearchServerTool, WebSearchServerTool$Outbound } from "./websearchservertool.js";
 import { WebSearchServerToolOpenRouter, WebSearchServerToolOpenRouter$Outbound } from "./websearchservertoolopenrouter.js";
-export type ResponsesRequestPlugin = AlignmentPlugin | AutoBetaRouterPlugin | AutoRouterPlugin | ContextCompressionPlugin | FileParserPlugin | FusionPlugin | JevRouterPlugin | ModerationPlugin | ParetoRouterPlugin | ResponseHealingPlugin | SwitchyardRouterPlugin | WebSearchPlugin | WebFetchPlugin;
+export type ResponsesRequestPlugin = AlignmentPlugin | ArmorPlugin | AutoBetaRouterPlugin | AutoRouterPlugin | ContextCompressionPlugin | FileParserPlugin | FusionPlugin | JevRouterPlugin | ModerationPlugin | ParetoRouterPlugin | ResponseHealingPlugin | SwitchyardRouterPlugin | WebSearchPlugin | WebFetchPlugin;
 export type ReasoningConfig = {
     /**
      * Controls which reasoning is available to the model. `auto` uses the model default (same as omitting); `all_turns` includes reasoning from earlier turns passed in input; `current_turn` limits to the current turn only. Only supported by OpenAI GPT-5.6 and newer.
@@ -89,6 +91,15 @@ export declare const ResponsesRequestServiceTier: {
  * The service tier to use for processing this request. `fast` is accepted as an alias for `priority`. `ultrafast` prefers ultrafast endpoints and falls back to `priority`, then default endpoints.
  */
 export type ResponsesRequestServiceTier = OpenEnum<typeof ResponsesRequestServiceTier>;
+/**
+ * Stop sequences (up to 4)
+ */
+export type ResponsesRequestStop = string | Array<string>;
+export declare const ResponsesRequestAllowedCaller: {
+    readonly Direct: "direct";
+    readonly Programmatic: "programmatic";
+};
+export type ResponsesRequestAllowedCaller = OpenEnum<typeof ResponsesRequestAllowedCaller>;
 /**
  * Function tool definition
  */
@@ -100,14 +111,18 @@ export type ResponsesRequestToolFunction = {
     } | null;
     strict?: boolean | null | undefined;
     type: "function";
+    allowedCallers?: Array<ResponsesRequestAllowedCaller> | null | undefined;
     /**
      * Lets the model keep working after calling this tool instead of waiting for its output. The tool is still executed by the client; return the result in a later request as a `function_call_output` with the original `call_id`. Only honored by providers whose Responses API supports async tools; ignored elsewhere.
      */
     async?: boolean | undefined;
     /**
-     * Withhold this tool from the model until `openrouter:tool_search` finds it. Requires the tool search server tool; at least one tool must remain non-deferred.
+     * Withhold this tool from the model until `openrouter:tool_search` finds it. Where the request declares no search tool, OpenRouter may add `openrouter:tool_search` to serve the flag, and otherwise sends the tool in full. A request that declares the search tool itself must keep at least one tool non-deferred.
      */
     deferLoading?: boolean | undefined;
+    outputSchema?: {
+        [k: string]: any;
+    } | null | undefined;
 };
 export type ResponsesRequestToolUnion = ResponsesRequestToolFunction | (PreviewWebSearchServerTool & {
     type: "web_search_preview";
@@ -177,6 +192,10 @@ export type ResponsesRequest = {
      * Debug options for inspecting request transformations (streaming only)
      */
     debug?: ChatDebugOptions | undefined;
+    /**
+     * Opt-in versioned router-level deferred-tool protocol. Replay assistant reasoning unchanged on continuation; keep the catalog unchanged.
+     */
+    deferredTools?: DeferredToolsControl | undefined;
     frequencyPenalty?: number | null | undefined;
     /**
      * Provider-specific image configuration options. Keys and values vary by model/provider. See https://openrouter.ai/docs/guides/overview/multimodal/image-generation for more details.
@@ -211,7 +230,7 @@ export type ResponsesRequest = {
     /**
      * Plugins you want to enable for this request, including their settings.
      */
-    plugins?: Array<AlignmentPlugin | AutoBetaRouterPlugin | AutoRouterPlugin | ContextCompressionPlugin | FileParserPlugin | FusionPlugin | JevRouterPlugin | ModerationPlugin | ParetoRouterPlugin | ResponseHealingPlugin | SwitchyardRouterPlugin | WebSearchPlugin | WebFetchPlugin> | undefined;
+    plugins?: Array<AlignmentPlugin | ArmorPlugin | AutoBetaRouterPlugin | AutoRouterPlugin | ContextCompressionPlugin | FileParserPlugin | FusionPlugin | JevRouterPlugin | ModerationPlugin | ParetoRouterPlugin | ResponseHealingPlugin | SwitchyardRouterPlugin | WebSearchPlugin | WebFetchPlugin> | undefined;
     presencePenalty?: number | null | undefined;
     /**
      * Not supported on this proxy. Each response request is independent: no responses are stored, so a previous response cannot be referenced. Requests with a non-null value are rejected with a 400 error. Send the full conversation history in `input` instead.
@@ -220,7 +239,7 @@ export type ResponsesRequest = {
     prompt?: StoredPromptTemplate | null | undefined;
     promptCacheKey?: string | null | undefined;
     /**
-     * Request-level prompt-cache controls. `mode: "explicit"` disables OpenAI-managed breakpoints so only blocks marked with `prompt_cache_breakpoint` are cached. Only supported by OpenAI GPT-5.6 and newer.
+     * Request-level prompt-cache controls. `mode: "explicit"` disables OpenAI-managed breakpoints so only blocks marked with `prompt_cache_breakpoint` are cached. `ttl` sets the minimum cache lifetime for every breakpoint the request writes: "30m" (default) or "12h" (billed at the higher 12h cache-write rate). Only supported by OpenAI GPT-5.6 and newer.
      */
     promptCacheOptions?: PromptCacheOptions | null | undefined;
     /**
@@ -243,6 +262,10 @@ export type ResponsesRequest = {
      * A unique identifier for grouping related requests (e.g., a conversation or agent workflow). When provided, OpenRouter uses it as the sticky routing key, routing all requests in the session to the same provider to maximize prompt cache hits. Also used for observability grouping. If provided in both the request body and the x-session-id header, the body value takes precedence. Maximum of 256 characters.
      */
     sessionId?: string | undefined;
+    /**
+     * Stop sequences (up to 4)
+     */
+    stop?: string | Array<string> | null | undefined;
     /**
      * Stop conditions for the server-tool agent loop. Any condition firing halts the loop (OR logic). When set, this overrides `max_tool_calls`. When a condition fires while the model is still emitting tool calls, the pending tool calls are executed and one final turn is made with tool calls disabled so the response ends with a natural-language answer instead of an unfinished tool call.
      */
@@ -324,7 +347,7 @@ export type ResponsesRequest = {
     user?: string | undefined;
 };
 /** @internal */
-export type ResponsesRequestPlugin$Outbound = AlignmentPlugin$Outbound | AutoBetaRouterPlugin$Outbound | AutoRouterPlugin$Outbound | ContextCompressionPlugin$Outbound | FileParserPlugin$Outbound | FusionPlugin$Outbound | JevRouterPlugin$Outbound | ModerationPlugin$Outbound | ParetoRouterPlugin$Outbound | ResponseHealingPlugin$Outbound | SwitchyardRouterPlugin$Outbound | WebSearchPlugin$Outbound | WebFetchPlugin$Outbound;
+export type ResponsesRequestPlugin$Outbound = AlignmentPlugin$Outbound | ArmorPlugin$Outbound | AutoBetaRouterPlugin$Outbound | AutoRouterPlugin$Outbound | ContextCompressionPlugin$Outbound | FileParserPlugin$Outbound | FusionPlugin$Outbound | JevRouterPlugin$Outbound | ModerationPlugin$Outbound | ParetoRouterPlugin$Outbound | ResponseHealingPlugin$Outbound | SwitchyardRouterPlugin$Outbound | WebSearchPlugin$Outbound | WebFetchPlugin$Outbound;
 /** @internal */
 export declare const ResponsesRequestPlugin$outboundSchema: z.ZodType<ResponsesRequestPlugin$Outbound, ResponsesRequestPlugin>;
 export declare function responsesRequestPluginToJSON(responsesRequestPlugin: ResponsesRequestPlugin): string;
@@ -343,6 +366,13 @@ export declare function reasoningConfigToJSON(reasoningConfig: ReasoningConfig):
 /** @internal */
 export declare const ResponsesRequestServiceTier$outboundSchema: z.ZodType<string, ResponsesRequestServiceTier>;
 /** @internal */
+export type ResponsesRequestStop$Outbound = string | Array<string>;
+/** @internal */
+export declare const ResponsesRequestStop$outboundSchema: z.ZodType<ResponsesRequestStop$Outbound, ResponsesRequestStop>;
+export declare function responsesRequestStopToJSON(responsesRequestStop: ResponsesRequestStop): string;
+/** @internal */
+export declare const ResponsesRequestAllowedCaller$outboundSchema: z.ZodType<string, ResponsesRequestAllowedCaller>;
+/** @internal */
 export type ResponsesRequestToolFunction$Outbound = {
     description?: string | null | undefined;
     name: string;
@@ -351,8 +381,12 @@ export type ResponsesRequestToolFunction$Outbound = {
     } | null;
     strict?: boolean | null | undefined;
     type: "function";
+    allowed_callers?: Array<string> | null | undefined;
     async?: boolean | undefined;
     defer_loading?: boolean | undefined;
+    output_schema?: {
+        [k: string]: any;
+    } | null | undefined;
 };
 /** @internal */
 export declare const ResponsesRequestToolFunction$outboundSchema: z.ZodType<ResponsesRequestToolFunction$Outbound, ResponsesRequestToolFunction>;
@@ -421,6 +455,7 @@ export type ResponsesRequest$Outbound = {
     background?: boolean | null | undefined;
     cache_control?: AnthropicCacheControlDirective$Outbound | undefined;
     debug?: ChatDebugOptions$Outbound | undefined;
+    deferred_tools?: DeferredToolsControl$Outbound | undefined;
     frequency_penalty?: number | null | undefined;
     image_config?: {
         [k: string]: ImageConfig$Outbound;
@@ -437,7 +472,7 @@ export type ResponsesRequest$Outbound = {
     model?: string | undefined;
     models?: Array<string> | undefined;
     parallel_tool_calls?: boolean | null | undefined;
-    plugins?: Array<AlignmentPlugin$Outbound | AutoBetaRouterPlugin$Outbound | AutoRouterPlugin$Outbound | ContextCompressionPlugin$Outbound | FileParserPlugin$Outbound | FusionPlugin$Outbound | JevRouterPlugin$Outbound | ModerationPlugin$Outbound | ParetoRouterPlugin$Outbound | ResponseHealingPlugin$Outbound | SwitchyardRouterPlugin$Outbound | WebSearchPlugin$Outbound | WebFetchPlugin$Outbound> | undefined;
+    plugins?: Array<AlignmentPlugin$Outbound | ArmorPlugin$Outbound | AutoBetaRouterPlugin$Outbound | AutoRouterPlugin$Outbound | ContextCompressionPlugin$Outbound | FileParserPlugin$Outbound | FusionPlugin$Outbound | JevRouterPlugin$Outbound | ModerationPlugin$Outbound | ParetoRouterPlugin$Outbound | ResponseHealingPlugin$Outbound | SwitchyardRouterPlugin$Outbound | WebSearchPlugin$Outbound | WebFetchPlugin$Outbound> | undefined;
     presence_penalty?: number | null | undefined;
     previous_response_id?: any | undefined;
     prompt?: StoredPromptTemplate$Outbound | null | undefined;
@@ -448,6 +483,7 @@ export type ResponsesRequest$Outbound = {
     safety_identifier?: string | null | undefined;
     service_tier: string | null;
     session_id?: string | undefined;
+    stop?: string | Array<string> | null | undefined;
     stop_server_tools_when?: Array<StopServerToolsWhenCondition$Outbound> | undefined;
     store: false;
     stream: boolean;