From ed1ea2690cd2e04ea0d8fcf57511c266df9991f2 Mon Sep 17 00:00:00 2001 From: Ramesh Vagadiya Date: Fri, 18 Apr 2025 12:41:13 +0100 Subject: [PATCH 1/2] Updates Claude 3.7 model token limits Increases maxTokens from 8192 to 64000 for Claude 3.7 Sonnet model configurations across various API integrations including Anthropic, Bedrock, Glama, Requesty, and OpenRouter. Also adjusts Vertex AI thinking model maxTokens to 128000 for improved performance. --- src/shared/api.ts | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/src/shared/api.ts b/src/shared/api.ts index 4d71d947ba..2c0482d003 100644 --- a/src/shared/api.ts +++ b/src/shared/api.ts @@ -27,7 +27,7 @@ export const anthropicModels = { thinking: true, }, "claude-3-7-sonnet-20250219": { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, @@ -166,7 +166,7 @@ export const bedrockModels = { cachableFields: ["system"], }, "anthropic.claude-3-7-sonnet-20250219-v1:0": { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, @@ -425,7 +425,7 @@ export const bedrockModels = { // https://glama.ai/models export const glamaDefaultModelId = "anthropic/claude-3-7-sonnet" export const glamaDefaultModelInfo: ModelInfo = { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, @@ -442,7 +442,7 @@ export const glamaDefaultModelInfo: ModelInfo = { // https://requesty.ai/router-2 export const requestyDefaultModelId = "anthropic/claude-3-7-sonnet-latest" export const requestyDefaultModelInfo: ModelInfo = { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, @@ -459,7 +459,7 @@ export const requestyDefaultModelInfo: ModelInfo = { // https://openrouter.ai/models?order=newest&supported_parameters=tools export const openRouterDefaultModelId = "anthropic/claude-3.7-sonnet" export const openRouterDefaultModelInfo: ModelInfo = { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, @@ -550,7 +550,7 @@ export const vertexModels = { outputPrice: 5, }, "claude-3-7-sonnet@20250219:thinking": { - maxTokens: 64_000, + maxTokens: 128_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, @@ -562,7 +562,7 @@ export const vertexModels = { thinking: true, }, "claude-3-7-sonnet@20250219": { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, From c3e22d53c33ac0cc6017d9c4714ca8dec444049c Mon Sep 17 00:00:00 2001 From: Ramesh Vagadiya Date: Fri, 18 Apr 2025 13:11:18 +0100 Subject: [PATCH 2/2] Updates max tokens limit to 64,000 across test files Increases maximum token limit from 8,192 to 64,000 in test expectations for multiple AI providers: - Anthropic - Glama - Requesty - Vertex This aligns test expectations with newer model capabilities that support larger token limits. --- src/api/providers/__tests__/anthropic.test.ts | 2 +- src/api/providers/__tests__/glama.test.ts | 4 ++-- src/api/providers/__tests__/requesty.test.ts | 2 +- src/api/providers/__tests__/vertex.test.ts | 4 ++-- 4 files changed, 6 insertions(+), 6 deletions(-) diff --git a/src/api/providers/__tests__/anthropic.test.ts b/src/api/providers/__tests__/anthropic.test.ts index fe186e3d8f..cb893e1e62 100644 --- a/src/api/providers/__tests__/anthropic.test.ts +++ b/src/api/providers/__tests__/anthropic.test.ts @@ -255,7 +255,7 @@ describe("AnthropicHandler", () => { }) const result = handler.getModel() - expect(result.maxTokens).toBe(8192) + expect(result.maxTokens).toBe(64_000) expect(result.thinking).toBeUndefined() expect(result.temperature).toBe(0) }) diff --git a/src/api/providers/__tests__/glama.test.ts b/src/api/providers/__tests__/glama.test.ts index 5e017ccd0a..a1666e21d3 100644 --- a/src/api/providers/__tests__/glama.test.ts +++ b/src/api/providers/__tests__/glama.test.ts @@ -181,7 +181,7 @@ describe("GlamaHandler", () => { model: mockOptions.apiModelId, messages: [{ role: "user", content: "Test prompt" }], temperature: 0, - max_tokens: 8192, + max_tokens: 64_000, }), ) }) @@ -233,7 +233,7 @@ describe("GlamaHandler", () => { const modelInfo = handler.getModel() expect(modelInfo.id).toBe(mockOptions.apiModelId) expect(modelInfo.info).toBeDefined() - expect(modelInfo.info.maxTokens).toBe(8192) + expect(modelInfo.info.maxTokens).toBe(64_000) expect(modelInfo.info.contextWindow).toBe(200_000) }) }) diff --git a/src/api/providers/__tests__/requesty.test.ts b/src/api/providers/__tests__/requesty.test.ts index 2b3da4a7ad..3b857dcd3d 100644 --- a/src/api/providers/__tests__/requesty.test.ts +++ b/src/api/providers/__tests__/requesty.test.ts @@ -18,7 +18,7 @@ describe("RequestyHandler", () => { requestyApiKey: "test-key", requestyModelId: "test-model", requestyModelInfo: { - maxTokens: 8192, + maxTokens: 64_000, contextWindow: 200_000, supportsImages: true, supportsComputerUse: true, diff --git a/src/api/providers/__tests__/vertex.test.ts b/src/api/providers/__tests__/vertex.test.ts index 6c4e891d0b..a865cdb01c 100644 --- a/src/api/providers/__tests__/vertex.test.ts +++ b/src/api/providers/__tests__/vertex.test.ts @@ -309,7 +309,7 @@ describe("VertexHandler", () => { }, ], generationConfig: { - maxOutputTokens: 8192, + maxOutputTokens: 64_000, temperature: 0, }, }) @@ -914,7 +914,7 @@ describe("VertexHandler", () => { }) const result = handler.getModel() - expect(result.maxTokens).toBe(8192) + expect(result.maxTokens).toBe(64_000) expect(result.thinking).toBeUndefined() expect(result.temperature).toBe(0) })