Merge pull request #1338 from RooVetGit/openrouter_think_big

Support 128k max tokens for openrouter and vertex thinking

Author: mrubens

src/api/providers/__tests__/openrouter.test.ts

@@ -72,7 +72,7 @@ describe("OpenRouterHandler", () => {
 			openRouterModelId: "test-model",
 			openRouterModelInfo: {
 				...mockOpenRouterModelInfo,
-				maxTokens: 64_000,
+				maxTokens: 128_000,
 				thinking: true,
 			},
 			modelMaxTokens: 32_768,

src/api/providers/openrouter.ts

@@ -263,7 +263,7 @@ export async function getOpenRouterModels() {
 					modelInfo.supportsPromptCache = true
 					modelInfo.cacheWritesPrice = 3.75
 					modelInfo.cacheReadsPrice = 0.3
-					modelInfo.maxTokens = rawModel.id === "anthropic/claude-3.7-sonnet:thinking" ? 64_000 : 16_384
+					modelInfo.maxTokens = rawModel.id === "anthropic/claude-3.7-sonnet:thinking" ? 128_000 : 16_384
 					break
 				case rawModel.id.startsWith("anthropic/claude-3.5-sonnet-20240620"):
 					modelInfo.supportsPromptCache = true

src/shared/api.ts

@@ -477,7 +477,7 @@ export const vertexModels = {
 		outputPrice: 5,
 	},
 	"claude-3-7-sonnet@20250219:thinking": {
-		maxTokens: 64_000,
+		maxTokens: 128_000,
 		contextWindow: 200_000,
 		supportsImages: true,
 		supportsComputerUse: true,