mirror of
https://github.com/danny-avila/LibreChat.git
synced 2025-09-22 06:00:56 +02:00
🧪 feat: Claude Sonnet 4 - 1M Context Window (Beta Header) (#9093)
* adding beta header context-1m-2025-08-07 to claude sonnet 4 to increase contact window to 1M tokens * adding context-1m beta header to test cases * ci: Update Anthropic `getLLMConfig` tests and headers for model variations - Refactored test cases to ensure proper handling of model variations for 'claude-sonnet-4'. - Cleaned up unused mock implementations in tests for better clarity and performance. * refactor: regex in header retrieval for 'claude-sonnet-4' models * refactor: default tokens for 'claude-sonnet-4' to `1,000,000` * refactor: add token value retrieval and pattern matching to model tests --------- Co-authored-by: Dirk Petersen <no-reply@nowhere.com>
This commit is contained in:
Danny Avila
GitHub
parent
a6d7ebf22e
commit
ad1503abdc
5 changed files with 105 additions and 104 deletions
|
|
@ -245,7 +245,7 @@ describe('AnthropicClient', () => {
|
|||
});
|
||||
|
||||
describe('Claude 4 model headers', () => {
|
||||
it('should add "prompt-caching" beta header for claude-sonnet-4 model', () => {
|
||||
it('should add "prompt-caching" and "context-1m" beta headers for claude-sonnet-4 model', () => {
|
||||
const client = new AnthropicClient('test-api-key');
|
||||
const modelOptions = {
|
||||
model: 'claude-sonnet-4-20250514',
|
||||
|
|
@ -255,10 +255,30 @@ describe('AnthropicClient', () => {
|
|||
expect(anthropicClient._options.defaultHeaders).toBeDefined();
|
||||
expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
|
||||
expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
|
||||
'prompt-caching-2024-07-31',
|
||||
'prompt-caching-2024-07-31,context-1m-2025-08-07',
|
||||
);
|
||||
});
|
||||
|
||||
it('should add "prompt-caching" and "context-1m" beta headers for claude-sonnet-4 model formats', () => {
|
||||
const client = new AnthropicClient('test-api-key');
|
||||
const modelVariations = [
|
||||
'claude-sonnet-4-20250514',
|
||||
'claude-sonnet-4-latest',
|
||||
'anthropic/claude-sonnet-4-20250514',
|
||||
];
|
||||
|
||||
modelVariations.forEach((model) => {
|
||||
const modelOptions = { model };
|
||||
client.setOptions({ modelOptions, promptCache: true });
|
||||
const anthropicClient = client.getClient(modelOptions);
|
||||
expect(anthropicClient._options.defaultHeaders).toBeDefined();
|
||||
expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
|
||||
expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
|
||||
'prompt-caching-2024-07-31,context-1m-2025-08-07',
|
||||
);
|
||||
});
|
||||
});
|
||||
|
||||
it('should add "prompt-caching" beta header for claude-opus-4 model', () => {
|
||||
const client = new AnthropicClient('test-api-key');
|
||||
const modelOptions = {
|
||||
|
|
@ -273,20 +293,6 @@ describe('AnthropicClient', () => {
|
|||
);
|
||||
});
|
||||
|
||||
it('should add "prompt-caching" beta header for claude-4-sonnet model', () => {
|
||||
const client = new AnthropicClient('test-api-key');
|
||||
const modelOptions = {
|
||||
model: 'claude-4-sonnet-20250514',
|
||||
};
|
||||
client.setOptions({ modelOptions, promptCache: true });
|
||||
const anthropicClient = client.getClient(modelOptions);
|
||||
expect(anthropicClient._options.defaultHeaders).toBeDefined();
|
||||
expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
|
||||
expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
|
||||
'prompt-caching-2024-07-31',
|
||||
);
|
||||
});
|
||||
|
||||
it('should add "prompt-caching" beta header for claude-4-opus model', () => {
|
||||
const client = new AnthropicClient('test-api-key');
|
||||
const modelOptions = {
|
||||
|
|
|
|||
|
|
@ -45,6 +45,10 @@ function getClaudeHeaders(model, supportsCacheControl) {
|
|||
'anthropic-beta':
|
||||
'token-efficient-tools-2025-02-19,output-128k-2025-02-19,prompt-caching-2024-07-31',
|
||||
};
|
||||
} else if (/claude-sonnet-4/.test(model)) {
|
||||
return {
|
||||
'anthropic-beta': 'prompt-caching-2024-07-31,context-1m-2025-08-07',
|
||||
};
|
||||
} else if (
|
||||
/claude-(?:sonnet|opus|haiku)-[4-9]/.test(model) ||
|
||||
/claude-[4-9]-(?:sonnet|opus|haiku)?/.test(model) ||
|
||||
|
|
|
|||
|
|
@ -1,50 +1,19 @@
|
|||
const { anthropicSettings, removeNullishValues } = require('librechat-data-provider');
|
||||
const { getLLMConfig } = require('~/server/services/Endpoints/anthropic/llm');
|
||||
const { checkPromptCacheSupport, getClaudeHeaders, configureReasoning } = require('./helpers');
|
||||
|
||||
jest.mock('https-proxy-agent', () => ({
|
||||
HttpsProxyAgent: jest.fn().mockImplementation((proxy) => ({ proxy })),
|
||||
}));
|
||||
|
||||
jest.mock('./helpers', () => ({
|
||||
checkPromptCacheSupport: jest.fn(),
|
||||
getClaudeHeaders: jest.fn(),
|
||||
configureReasoning: jest.fn((requestOptions) => requestOptions),
|
||||
}));
|
||||
|
||||
jest.mock('librechat-data-provider', () => ({
|
||||
anthropicSettings: {
|
||||
model: { default: 'claude-3-opus-20240229' },
|
||||
maxOutputTokens: { default: 4096, reset: jest.fn(() => 4096) },
|
||||
thinking: { default: false },
|
||||
promptCache: { default: false },
|
||||
thinkingBudget: { default: null },
|
||||
},
|
||||
removeNullishValues: jest.fn((obj) => {
|
||||
const result = {};
|
||||
for (const key in obj) {
|
||||
if (obj[key] !== null && obj[key] !== undefined) {
|
||||
result[key] = obj[key];
|
||||
}
|
||||
}
|
||||
return result;
|
||||
}),
|
||||
}));
|
||||
|
||||
describe('getLLMConfig', () => {
|
||||
beforeEach(() => {
|
||||
jest.clearAllMocks();
|
||||
checkPromptCacheSupport.mockReturnValue(false);
|
||||
getClaudeHeaders.mockReturnValue(undefined);
|
||||
configureReasoning.mockImplementation((requestOptions) => requestOptions);
|
||||
anthropicSettings.maxOutputTokens.reset.mockReturnValue(4096);
|
||||
});
|
||||
|
||||
it('should create a basic configuration with default values', () => {
|
||||
const result = getLLMConfig('test-api-key', { modelOptions: {} });
|
||||
|
||||
expect(result.llmConfig).toHaveProperty('apiKey', 'test-api-key');
|
||||
expect(result.llmConfig).toHaveProperty('model', anthropicSettings.model.default);
|
||||
expect(result.llmConfig).toHaveProperty('model', 'claude-3-5-sonnet-latest');
|
||||
expect(result.llmConfig).toHaveProperty('stream', true);
|
||||
expect(result.llmConfig).toHaveProperty('maxTokens');
|
||||
});
|
||||
|
|
@ -99,40 +68,73 @@ describe('getLLMConfig', () => {
|
|||
expect(result.llmConfig).toHaveProperty('topP', 0.9);
|
||||
});
|
||||
|
||||
it('should NOT include topK and topP for Claude-3-7 models (hyphen notation)', () => {
|
||||
configureReasoning.mockImplementation((requestOptions) => {
|
||||
requestOptions.thinking = { type: 'enabled' };
|
||||
return requestOptions;
|
||||
});
|
||||
|
||||
it('should NOT include topK and topP for Claude-3-7 models with thinking enabled (hyphen notation)', () => {
|
||||
const result = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
model: 'claude-3-7-sonnet',
|
||||
topK: 10,
|
||||
topP: 0.9,
|
||||
thinking: true,
|
||||
},
|
||||
});
|
||||
|
||||
expect(result.llmConfig).not.toHaveProperty('topK');
|
||||
expect(result.llmConfig).not.toHaveProperty('topP');
|
||||
expect(result.llmConfig).toHaveProperty('thinking');
|
||||
expect(result.llmConfig.thinking).toHaveProperty('type', 'enabled');
|
||||
// When thinking is enabled, it uses the default thinkingBudget of 2000
|
||||
expect(result.llmConfig.thinking).toHaveProperty('budget_tokens', 2000);
|
||||
});
|
||||
|
||||
it('should NOT include topK and topP for Claude-3.7 models (decimal notation)', () => {
|
||||
configureReasoning.mockImplementation((requestOptions) => {
|
||||
requestOptions.thinking = { type: 'enabled' };
|
||||
return requestOptions;
|
||||
it('should add "prompt-caching" and "context-1m" beta headers for claude-sonnet-4 model', () => {
|
||||
const modelOptions = {
|
||||
model: 'claude-sonnet-4-20250514',
|
||||
promptCache: true,
|
||||
};
|
||||
const result = getLLMConfig('test-key', { modelOptions });
|
||||
const clientOptions = result.llmConfig.clientOptions;
|
||||
expect(clientOptions.defaultHeaders).toBeDefined();
|
||||
expect(clientOptions.defaultHeaders).toHaveProperty('anthropic-beta');
|
||||
expect(clientOptions.defaultHeaders['anthropic-beta']).toBe(
|
||||
'prompt-caching-2024-07-31,context-1m-2025-08-07',
|
||||
);
|
||||
});
|
||||
|
||||
it('should add "prompt-caching" and "context-1m" beta headers for claude-sonnet-4 model formats', () => {
|
||||
const modelVariations = [
|
||||
'claude-sonnet-4-20250514',
|
||||
'claude-sonnet-4-latest',
|
||||
'anthropic/claude-sonnet-4-20250514',
|
||||
];
|
||||
|
||||
modelVariations.forEach((model) => {
|
||||
const modelOptions = { model, promptCache: true };
|
||||
const result = getLLMConfig('test-key', { modelOptions });
|
||||
const clientOptions = result.llmConfig.clientOptions;
|
||||
expect(clientOptions.defaultHeaders).toBeDefined();
|
||||
expect(clientOptions.defaultHeaders).toHaveProperty('anthropic-beta');
|
||||
expect(clientOptions.defaultHeaders['anthropic-beta']).toBe(
|
||||
'prompt-caching-2024-07-31,context-1m-2025-08-07',
|
||||
);
|
||||
});
|
||||
});
|
||||
|
||||
it('should NOT include topK and topP for Claude-3.7 models with thinking enabled (decimal notation)', () => {
|
||||
const result = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
model: 'claude-3.7-sonnet',
|
||||
topK: 10,
|
||||
topP: 0.9,
|
||||
thinking: true,
|
||||
},
|
||||
});
|
||||
|
||||
expect(result.llmConfig).not.toHaveProperty('topK');
|
||||
expect(result.llmConfig).not.toHaveProperty('topP');
|
||||
expect(result.llmConfig).toHaveProperty('thinking');
|
||||
expect(result.llmConfig.thinking).toHaveProperty('type', 'enabled');
|
||||
// When thinking is enabled, it uses the default thinkingBudget of 2000
|
||||
expect(result.llmConfig.thinking).toHaveProperty('budget_tokens', 2000);
|
||||
});
|
||||
|
||||
it('should handle custom maxOutputTokens', () => {
|
||||
|
|
@ -233,7 +235,6 @@ describe('getLLMConfig', () => {
|
|||
});
|
||||
|
||||
it('should handle maxOutputTokens when explicitly set to falsy value', () => {
|
||||
anthropicSettings.maxOutputTokens.reset.mockReturnValue(8192);
|
||||
const result = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
model: 'claude-3-opus',
|
||||
|
|
@ -241,8 +242,8 @@ describe('getLLMConfig', () => {
|
|||
},
|
||||
});
|
||||
|
||||
expect(anthropicSettings.maxOutputTokens.reset).toHaveBeenCalledWith('claude-3-opus');
|
||||
expect(result.llmConfig).toHaveProperty('maxTokens', 8192);
|
||||
// The actual anthropicSettings.maxOutputTokens.reset('claude-3-opus') returns 4096
|
||||
expect(result.llmConfig).toHaveProperty('maxTokens', 4096);
|
||||
});
|
||||
|
||||
it('should handle both proxy and reverseProxyUrl', () => {
|
||||
|
|
@ -263,9 +264,6 @@ describe('getLLMConfig', () => {
|
|||
});
|
||||
|
||||
it('should handle prompt cache with supported model', () => {
|
||||
checkPromptCacheSupport.mockReturnValue(true);
|
||||
getClaudeHeaders.mockReturnValue({ 'anthropic-beta': 'prompt-caching-2024-07-31' });
|
||||
|
||||
const result = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
model: 'claude-3-5-sonnet',
|
||||
|
|
@ -273,43 +271,38 @@ describe('getLLMConfig', () => {
|
|||
},
|
||||
});
|
||||
|
||||
expect(checkPromptCacheSupport).toHaveBeenCalledWith('claude-3-5-sonnet');
|
||||
expect(getClaudeHeaders).toHaveBeenCalledWith('claude-3-5-sonnet', true);
|
||||
// claude-3-5-sonnet supports prompt caching and should get the appropriate headers
|
||||
expect(result.llmConfig.clientOptions.defaultHeaders).toEqual({
|
||||
'anthropic-beta': 'prompt-caching-2024-07-31',
|
||||
'anthropic-beta': 'max-tokens-3-5-sonnet-2024-07-15,prompt-caching-2024-07-31',
|
||||
});
|
||||
});
|
||||
|
||||
it('should handle thinking and thinkingBudget options', () => {
|
||||
configureReasoning.mockImplementation((requestOptions, systemOptions) => {
|
||||
if (systemOptions.thinking) {
|
||||
requestOptions.thinking = { type: 'enabled' };
|
||||
}
|
||||
if (systemOptions.thinkingBudget) {
|
||||
requestOptions.thinking = {
|
||||
...requestOptions.thinking,
|
||||
budget_tokens: systemOptions.thinkingBudget,
|
||||
};
|
||||
}
|
||||
return requestOptions;
|
||||
});
|
||||
|
||||
getLLMConfig('test-api-key', {
|
||||
const result = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
model: 'claude-3-7-sonnet',
|
||||
thinking: true,
|
||||
thinkingBudget: 5000,
|
||||
thinkingBudget: 10000, // This exceeds the default max_tokens of 8192
|
||||
},
|
||||
});
|
||||
|
||||
expect(configureReasoning).toHaveBeenCalledWith(
|
||||
expect.any(Object),
|
||||
expect.objectContaining({
|
||||
// The function should add thinking configuration for claude-3-7 models
|
||||
expect(result.llmConfig).toHaveProperty('thinking');
|
||||
expect(result.llmConfig.thinking).toHaveProperty('type', 'enabled');
|
||||
// With claude-3-7-sonnet, the max_tokens default is 8192
|
||||
// Budget tokens gets adjusted to 90% of max_tokens (8192 * 0.9 = 7372) when it exceeds max_tokens
|
||||
expect(result.llmConfig.thinking).toHaveProperty('budget_tokens', 7372);
|
||||
|
||||
// Test with budget_tokens within max_tokens limit
|
||||
const result2 = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
model: 'claude-3-7-sonnet',
|
||||
thinking: true,
|
||||
promptCache: false,
|
||||
thinkingBudget: 5000,
|
||||
}),
|
||||
);
|
||||
thinkingBudget: 2000,
|
||||
},
|
||||
});
|
||||
|
||||
expect(result2.llmConfig.thinking).toHaveProperty('budget_tokens', 2000);
|
||||
});
|
||||
|
||||
it('should remove system options from modelOptions', () => {
|
||||
|
|
@ -330,16 +323,6 @@ describe('getLLMConfig', () => {
|
|||
});
|
||||
|
||||
it('should handle all nullish values removal', () => {
|
||||
removeNullishValues.mockImplementation((obj) => {
|
||||
const cleaned = {};
|
||||
Object.entries(obj).forEach(([key, value]) => {
|
||||
if (value !== null && value !== undefined) {
|
||||
cleaned[key] = value;
|
||||
}
|
||||
});
|
||||
return cleaned;
|
||||
});
|
||||
|
||||
const result = getLLMConfig('test-api-key', {
|
||||
modelOptions: {
|
||||
temperature: null,
|
||||
|
|
|
|||
|
|
@ -108,7 +108,7 @@ const anthropicModels = {
|
|||
'claude-3.7-sonnet': 200000,
|
||||
'claude-3-5-sonnet-latest': 200000,
|
||||
'claude-3.5-sonnet-latest': 200000,
|
||||
'claude-sonnet-4': 200000,
|
||||
'claude-sonnet-4': 1000000,
|
||||
'claude-opus-4': 200000,
|
||||
'claude-4': 200000,
|
||||
};
|
||||
|
|
@ -487,5 +487,7 @@ module.exports = {
|
|||
matchModelName,
|
||||
processModelData,
|
||||
getModelMaxTokens,
|
||||
getModelTokenValue,
|
||||
findMatchingPattern,
|
||||
getModelMaxOutputTokens,
|
||||
};
|
||||
|
|
|
|||
|
|
@ -1,6 +1,7 @@
|
|||
const { EModelEndpoint } = require('librechat-data-provider');
|
||||
const {
|
||||
maxOutputTokensMap,
|
||||
findMatchingPattern,
|
||||
getModelMaxTokens,
|
||||
processModelData,
|
||||
matchModelName,
|
||||
|
|
@ -749,8 +750,12 @@ describe('Grok Model Tests - Tokens', () => {
|
|||
|
||||
describe('Claude Model Tests', () => {
|
||||
it('should return correct context length for Claude 4 models', () => {
|
||||
expect(getModelMaxTokens('claude-sonnet-4')).toBe(200000);
|
||||
expect(getModelMaxTokens('claude-opus-4')).toBe(200000);
|
||||
expect(getModelMaxTokens('claude-sonnet-4')).toBe(
|
||||
maxTokensMap[EModelEndpoint.anthropic]['claude-sonnet-4'],
|
||||
);
|
||||
expect(getModelMaxTokens('claude-opus-4')).toBe(
|
||||
maxTokensMap[EModelEndpoint.anthropic]['claude-opus-4'],
|
||||
);
|
||||
});
|
||||
|
||||
it('should handle Claude 4 model name variations with different prefixes and suffixes', () => {
|
||||
|
|
@ -772,7 +777,8 @@ describe('Claude Model Tests', () => {
|
|||
];
|
||||
|
||||
modelVariations.forEach((model) => {
|
||||
expect(getModelMaxTokens(model)).toBe(200000);
|
||||
const modelKey = findMatchingPattern(model, maxTokensMap[EModelEndpoint.anthropic]);
|
||||
expect(getModelMaxTokens(model)).toBe(maxTokensMap[EModelEndpoint.anthropic][modelKey]);
|
||||
});
|
||||
});
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue