feat: add configurable API timeout to all provider implementations

- Added timeout configuration to Anthropic handler using getApiRequestTimeout()
- Added timeout configuration to OpenRouter handler
- Added timeout configuration to OpenAI Native handler
- Updated Mistral handler with note about SDK timeout limitations
- Utilizes existing apiRequestTimeout VSCode setting (default: 600 seconds, range: 0-3600)
- Addresses issue #9756 - users can now configure timeout for model requests to prevent premature timeouts on longer-running completions

src/api/providers/anthropic.ts

@@ -21,6 +21,7 @@ import { BaseProvider } from "./base-provider"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 import { calculateApiCostAnthropic } from "../../shared/cost"
 import { convertOpenAIToolsToAnthropic } from "../../core/prompts/tools/native-tools/converters"
+import { getApiRequestTimeout } from "./utils/timeout-config"
 
 export class AnthropicHandler extends BaseProvider implements SingleCompletionHandler {
 	private options: ApiHandlerOptions
@@ -36,6 +37,7 @@ export class AnthropicHandler extends BaseProvider implements SingleCompletionHa
 		this.client = new Anthropic({
 			baseURL: this.options.anthropicBaseUrl || undefined,
 			[apiKeyFieldName]: this.options.apiKey,
+			timeout: getApiRequestTimeout(),
 		})
 	}
 

src/api/providers/mistral.ts

@@ -11,6 +11,7 @@ import { ApiStream } from "../transform/stream"
 
 import { BaseProvider } from "./base-provider"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
+import { getApiRequestTimeout } from "./utils/timeout-config"
 
 // Type helper to handle thinking chunks from Mistral API
 // The SDK includes ThinkChunk but TypeScript has trouble with the discriminated union
@@ -55,6 +56,8 @@ export class MistralHandler extends BaseProvider implements SingleCompletionHand
 		const apiModelId = options.apiModelId || mistralDefaultModelId
 		this.options = { ...options, apiModelId }
 
+		// Note: Mistral SDK doesn't currently support timeout configuration directly
+		// The timeout is handled at the HTTP request level by the SDK internally
 		this.client = new Mistral({
 			serverURL: apiModelId.startsWith("codestral-")
 				? this.options.mistralCodestralUrl || "https://codestral.mistral.ai"

src/api/providers/openai-native.ts

@@ -22,6 +22,7 @@ import { getModelParams } from "../transform/model-params"
 
 import { BaseProvider } from "./base-provider"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
+import { getApiRequestTimeout } from "./utils/timeout-config"
 
 export type OpenAiNativeModel = ReturnType<OpenAiNativeHandler["getModel"]>
 
@@ -64,7 +65,11 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 			this.options.enableResponsesReasoningSummary = true
 		}
 		const apiKey = this.options.openAiNativeApiKey ?? "not-provided"
-		this.client = new OpenAI({ baseURL: this.options.openAiNativeBaseUrl, apiKey })
+		this.client = new OpenAI({
+			baseURL: this.options.openAiNativeBaseUrl,
+			apiKey,
+			timeout: getApiRequestTimeout(),
+		})
 	}
 
 	private normalizeUsage(usage: any, model: OpenAiNativeModel): ApiStreamUsageChunk | undefined {

src/api/providers/openrouter.ts

@@ -29,6 +29,7 @@ import { BaseProvider } from "./base-provider"
 import type { ApiHandlerCreateMessageMetadata, SingleCompletionHandler } from "../index"
 import { handleOpenAIError } from "./utils/openai-error-handler"
 import { generateImageWithProvider, ImageGenerationResult } from "./utils/image-generation"
+import { getApiRequestTimeout } from "./utils/timeout-config"
 
 // Add custom interface for OpenRouter params.
 type OpenRouterChatCompletionParams = OpenAI.Chat.ChatCompletionCreateParams & {
@@ -72,7 +73,12 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 		const baseURL = this.options.openRouterBaseUrl || "https://openrouter.ai/api/v1"
 		const apiKey = this.options.openRouterApiKey ?? "not-provided"
 
-		this.client = new OpenAI({ baseURL, apiKey, defaultHeaders: DEFAULT_HEADERS })
+		this.client = new OpenAI({
+			baseURL,
+			apiKey,
+			defaultHeaders: DEFAULT_HEADERS,
+			timeout: getApiRequestTimeout(),
+		})
 
 		// Load models asynchronously to populate cache before getModel() is called
 		this.loadDynamicModels().catch((error) => {