🤖 feat: Claude Opus 4.5 Token Rates and Window Limits (#10653)
Some checks are pending
Docker Dev Branch Images Build / build (Dockerfile, lc-dev, node) (push) Waiting to run
Docker Dev Branch Images Build / build (Dockerfile.multi, lc-dev-api, api-build) (push) Waiting to run
Docker Dev Images Build / build (Dockerfile, librechat-dev, node) (push) Waiting to run
Docker Dev Images Build / build (Dockerfile.multi, librechat-dev-api, api-build) (push) Waiting to run
Sync Locize Translations & Create Translation PR / Sync Translation Keys with Locize (push) Waiting to run
Sync Locize Translations & Create Translation PR / Create Translation PR on Version Published (push) Blocked by required conditions

* 🤖 feat: Claude Opus 4.5 Token Rates and Window Limits

- Introduced new model 'claude-opus-4-5' with defined prompt and completion values in tokenValues and cacheTokenValues.
- Updated tests to validate prompt, completion, and cache rates for the new model.
- Enhanced model name handling to accommodate variations for 'claude-opus-4-5' across different contexts.
- Adjusted schemas to ensure correct max output token limits for the new model.

* ci: Add tests for "prompt-caching" beta header in Claude Opus 4.5 models

- Implemented tests to verify the addition of the "prompt-caching" beta header for the 'claude-opus-4-5' model and its variations.
- Updated future-proofing logic to ensure correct max token limits for Claude 4.x and 5.x Opus models, adjusting defaults to 64K where applicable.
- Enhanced existing tests to reflect changes in expected max token values for future Claude models.

* chore: Remove redundant max output check for Anthropic settings

- Eliminated the unnecessary check for ANTHROPIC_MAX_OUTPUT in the anthropicSettings schema, streamlining the logic for handling max output values.
This commit is contained in:
Danny Avila 2025-11-24 16:30:56 -05:00 committed by GitHub
parent e123e5f9ec
commit 9211d59388
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
8 changed files with 508 additions and 14 deletions

View file

@ -122,6 +122,38 @@ describe('getLLMConfig', () => {
});
});
it('should add "prompt-caching" beta header for claude-opus-4-5 model', () => {
const modelOptions = {
model: 'claude-opus-4-5',
promptCache: true,
};
const result = getLLMConfig('test-key', { modelOptions });
const clientOptions = result.llmConfig.clientOptions;
expect(clientOptions?.defaultHeaders).toBeDefined();
expect(clientOptions?.defaultHeaders).toHaveProperty('anthropic-beta');
const defaultHeaders = clientOptions?.defaultHeaders as Record<string, string>;
expect(defaultHeaders['anthropic-beta']).toBe('prompt-caching-2024-07-31');
});
it('should add "prompt-caching" beta header for claude-opus-4-5 model formats', () => {
const modelVariations = [
'claude-opus-4-5',
'claude-opus-4-5-20250420',
'claude-opus-4.5',
'anthropic/claude-opus-4-5',
];
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');
const defaultHeaders = clientOptions?.defaultHeaders as Record<string, string>;
expect(defaultHeaders['anthropic-beta']).toBe('prompt-caching-2024-07-31');
});
});
it('should NOT include topK and topP for Claude-3.7 models with thinking enabled (decimal notation)', () => {
const result = getLLMConfig('test-api-key', {
modelOptions: {
@ -707,6 +739,7 @@ describe('getLLMConfig', () => {
{ model: 'claude-haiku-4-5-20251001', expectedMaxTokens: 64000 },
{ model: 'claude-opus-4-1', expectedMaxTokens: 32000 },
{ model: 'claude-opus-4-1-20250805', expectedMaxTokens: 32000 },
{ model: 'claude-opus-4-5', expectedMaxTokens: 64000 },
{ model: 'claude-sonnet-4-20250514', expectedMaxTokens: 64000 },
{ model: 'claude-opus-4-0', expectedMaxTokens: 32000 },
];
@ -771,6 +804,17 @@ describe('getLLMConfig', () => {
});
});
it('should default Claude Opus 4.5 model to 64K tokens', () => {
const testCases = ['claude-opus-4-5', 'claude-opus-4-5-20250420', 'claude-opus-4.5'];
testCases.forEach((model) => {
const result = getLLMConfig('test-key', {
modelOptions: { model },
});
expect(result.llmConfig.maxTokens).toBe(64000);
});
});
it('should default future Claude 4.x Sonnet/Haiku models to 64K (future-proofing)', () => {
const testCases = ['claude-sonnet-4-20250514', 'claude-sonnet-4-9', 'claude-haiku-4-8'];
@ -782,15 +826,24 @@ describe('getLLMConfig', () => {
});
});
it('should default future Claude 4.x Opus models to 32K (future-proofing)', () => {
const testCases = ['claude-opus-4-0', 'claude-opus-4-7'];
testCases.forEach((model) => {
it('should default future Claude 4.x Opus models (future-proofing)', () => {
// opus-4-0 through opus-4-4 get 32K
const opus32kModels = ['claude-opus-4-0', 'claude-opus-4-1', 'claude-opus-4-4'];
opus32kModels.forEach((model) => {
const result = getLLMConfig('test-key', {
modelOptions: { model },
});
expect(result.llmConfig.maxTokens).toBe(32000);
});
// opus-4-5+ get 64K
const opus64kModels = ['claude-opus-4-5', 'claude-opus-4-7', 'claude-opus-4-10'];
opus64kModels.forEach((model) => {
const result = getLLMConfig('test-key', {
modelOptions: { model },
});
expect(result.llmConfig.maxTokens).toBe(64000);
});
});
it('should handle explicit maxOutputTokens override for Claude 4.x models', () => {
@ -908,7 +961,7 @@ describe('getLLMConfig', () => {
});
});
it('should future-proof Claude 5.x Opus models with 32K default', () => {
it('should future-proof Claude 5.x Opus models with 64K default', () => {
const testCases = [
'claude-opus-5',
'claude-opus-5-0',
@ -920,28 +973,28 @@ describe('getLLMConfig', () => {
const result = getLLMConfig('test-key', {
modelOptions: { model },
});
expect(result.llmConfig.maxTokens).toBe(32000);
expect(result.llmConfig.maxTokens).toBe(64000);
});
});
it('should future-proof Claude 6-9.x models with correct defaults', () => {
const testCases = [
// Claude 6.x
// Claude 6.x - All get 64K since they're version 5+
{ model: 'claude-sonnet-6', expected: 64000 },
{ model: 'claude-haiku-6-0', expected: 64000 },
{ model: 'claude-opus-6-1', expected: 32000 },
{ model: 'claude-opus-6-1', expected: 64000 }, // opus 6+ gets 64K
// Claude 7.x
{ model: 'claude-sonnet-7-20270101', expected: 64000 },
{ model: 'claude-haiku-7.5', expected: 64000 },
{ model: 'claude-opus-7', expected: 32000 },
{ model: 'claude-opus-7', expected: 64000 }, // opus 7+ gets 64K
// Claude 8.x
{ model: 'claude-sonnet-8', expected: 64000 },
{ model: 'claude-haiku-8-2', expected: 64000 },
{ model: 'claude-opus-8-latest', expected: 32000 },
{ model: 'claude-opus-8-latest', expected: 64000 }, // opus 8+ gets 64K
// Claude 9.x
{ model: 'claude-sonnet-9', expected: 64000 },
{ model: 'claude-haiku-9', expected: 64000 },
{ model: 'claude-opus-9', expected: 32000 },
{ model: 'claude-opus-9', expected: 64000 }, // opus 9+ gets 64K
];
testCases.forEach(({ model, expected }) => {

View file

@ -133,8 +133,9 @@ const anthropicModels = {
'claude-3.5-sonnet-latest': 200000,
'claude-haiku-4-5': 200000,
'claude-sonnet-4': 1000000,
'claude-opus-4': 200000,
'claude-4': 200000,
'claude-opus-4': 200000,
'claude-opus-4-5': 200000,
};
const deepseekModels = {
@ -334,8 +335,9 @@ const anthropicMaxOutputs = {
'claude-3-sonnet': 4096,
'claude-3-opus': 4096,
'claude-haiku-4-5': 64000,
'claude-opus-4': 32000,
'claude-sonnet-4': 64000,
'claude-opus-4': 32000,
'claude-opus-4-5': 64000,
'claude-3.5-sonnet': 8192,
'claude-3-5-sonnet': 8192,
'claude-3.7-sonnet': 128000,

View file

@ -1003,6 +1003,7 @@ const sharedAnthropicModels = [
'claude-haiku-4-5-20251001',
'claude-opus-4-1',
'claude-opus-4-1-20250805',
'claude-opus-4-5',
'claude-sonnet-4-20250514',
'claude-sonnet-4-0',
'claude-opus-4-20250514',

View file

@ -0,0 +1,341 @@
import { anthropicSettings } from './schemas';
describe('anthropicSettings', () => {
describe('maxOutputTokens.reset()', () => {
const { reset } = anthropicSettings.maxOutputTokens;
describe('Claude Sonnet models', () => {
it('should return 64K for claude-sonnet-4', () => {
expect(reset('claude-sonnet-4')).toBe(64000);
});
it('should return 64K for claude-sonnet-4-5', () => {
expect(reset('claude-sonnet-4-5')).toBe(64000);
});
it('should return 64K for claude-sonnet-5', () => {
expect(reset('claude-sonnet-5')).toBe(64000);
});
it('should return 64K for future versions like claude-sonnet-9', () => {
expect(reset('claude-sonnet-9')).toBe(64000);
});
});
describe('Claude Haiku models', () => {
it('should return 64K for claude-haiku-4-5', () => {
expect(reset('claude-haiku-4-5')).toBe(64000);
});
it('should return 64K for claude-haiku-4', () => {
expect(reset('claude-haiku-4')).toBe(64000);
});
it('should return 64K for claude-haiku-5', () => {
expect(reset('claude-haiku-5')).toBe(64000);
});
it('should return 64K for future versions like claude-haiku-9', () => {
expect(reset('claude-haiku-9')).toBe(64000);
});
});
describe('Claude Opus 4.0-4.4 models (32K limit)', () => {
it('should return 32K for claude-opus-4', () => {
expect(reset('claude-opus-4')).toBe(32000);
});
it('should return 32K for claude-opus-4-0', () => {
expect(reset('claude-opus-4-0')).toBe(32000);
});
it('should return 32K for claude-opus-4-1', () => {
expect(reset('claude-opus-4-1')).toBe(32000);
});
it('should return 32K for claude-opus-4-2', () => {
expect(reset('claude-opus-4-2')).toBe(32000);
});
it('should return 32K for claude-opus-4-3', () => {
expect(reset('claude-opus-4-3')).toBe(32000);
});
it('should return 32K for claude-opus-4-4', () => {
expect(reset('claude-opus-4-4')).toBe(32000);
});
it('should return 32K for claude-opus-4.0', () => {
expect(reset('claude-opus-4.0')).toBe(32000);
});
it('should return 32K for claude-opus-4.1', () => {
expect(reset('claude-opus-4.1')).toBe(32000);
});
});
describe('Claude Opus 4.5+ models (64K limit - future-proof)', () => {
it('should return 64K for claude-opus-4-5', () => {
expect(reset('claude-opus-4-5')).toBe(64000);
});
it('should return 64K for claude-opus-4-6', () => {
expect(reset('claude-opus-4-6')).toBe(64000);
});
it('should return 64K for claude-opus-4-7', () => {
expect(reset('claude-opus-4-7')).toBe(64000);
});
it('should return 64K for claude-opus-4-8', () => {
expect(reset('claude-opus-4-8')).toBe(64000);
});
it('should return 64K for claude-opus-4-9', () => {
expect(reset('claude-opus-4-9')).toBe(64000);
});
it('should return 64K for claude-opus-4.5', () => {
expect(reset('claude-opus-4.5')).toBe(64000);
});
it('should return 64K for claude-opus-4.6', () => {
expect(reset('claude-opus-4.6')).toBe(64000);
});
});
describe('Claude Opus 4.10+ models (double-digit minor versions)', () => {
it('should return 64K for claude-opus-4-10', () => {
expect(reset('claude-opus-4-10')).toBe(64000);
});
it('should return 64K for claude-opus-4-11', () => {
expect(reset('claude-opus-4-11')).toBe(64000);
});
it('should return 64K for claude-opus-4-15', () => {
expect(reset('claude-opus-4-15')).toBe(64000);
});
it('should return 64K for claude-opus-4-20', () => {
expect(reset('claude-opus-4-20')).toBe(64000);
});
it('should return 64K for claude-opus-4.10', () => {
expect(reset('claude-opus-4.10')).toBe(64000);
});
});
describe('Claude Opus 5+ models (future major versions)', () => {
it('should return 64K for claude-opus-5', () => {
expect(reset('claude-opus-5')).toBe(64000);
});
it('should return 64K for claude-opus-6', () => {
expect(reset('claude-opus-6')).toBe(64000);
});
it('should return 64K for claude-opus-7', () => {
expect(reset('claude-opus-7')).toBe(64000);
});
it('should return 64K for claude-opus-9', () => {
expect(reset('claude-opus-9')).toBe(64000);
});
it('should return 64K for claude-opus-5-0', () => {
expect(reset('claude-opus-5-0')).toBe(64000);
});
it('should return 64K for claude-opus-5.0', () => {
expect(reset('claude-opus-5.0')).toBe(64000);
});
});
describe('Model name variations with dates and suffixes', () => {
it('should return 64K for claude-opus-4-5-20250420', () => {
expect(reset('claude-opus-4-5-20250420')).toBe(64000);
});
it('should return 64K for claude-opus-4-6-20260101', () => {
expect(reset('claude-opus-4-6-20260101')).toBe(64000);
});
it('should return 32K for claude-opus-4-1-20250805', () => {
expect(reset('claude-opus-4-1-20250805')).toBe(32000);
});
it('should return 32K for claude-opus-4-0-20240229', () => {
expect(reset('claude-opus-4-0-20240229')).toBe(32000);
});
});
describe('Legacy Claude models', () => {
it('should return 8192 for claude-3-opus', () => {
expect(reset('claude-3-opus')).toBe(8192);
});
it('should return 8192 for claude-3-5-sonnet', () => {
expect(reset('claude-3-5-sonnet')).toBe(8192);
});
it('should return 8192 for claude-3-5-haiku', () => {
expect(reset('claude-3-5-haiku')).toBe(8192);
});
it('should return 8192 for claude-3-7-sonnet', () => {
expect(reset('claude-3-7-sonnet')).toBe(8192);
});
it('should return 8192 for claude-2', () => {
expect(reset('claude-2')).toBe(8192);
});
it('should return 8192 for claude-2.1', () => {
expect(reset('claude-2.1')).toBe(8192);
});
it('should return 8192 for claude-instant', () => {
expect(reset('claude-instant')).toBe(8192);
});
});
describe('Non-Claude models and edge cases', () => {
it('should return 8192 for unknown model', () => {
expect(reset('unknown-model')).toBe(8192);
});
it('should return 8192 for empty string', () => {
expect(reset('')).toBe(8192);
});
it('should return 8192 for gpt-4', () => {
expect(reset('gpt-4')).toBe(8192);
});
it('should return 8192 for gemini-pro', () => {
expect(reset('gemini-pro')).toBe(8192);
});
});
describe('Regex pattern edge cases', () => {
it('should not match claude-opus-3', () => {
expect(reset('claude-opus-3')).toBe(8192);
});
it('should not match opus-4-5 without claude prefix', () => {
expect(reset('opus-4-5')).toBe(8192);
});
it('should NOT match claude.opus.4.5 (incorrect separator pattern)', () => {
// Model names use hyphens after "claude", not dots
expect(reset('claude.opus.4.5')).toBe(8192);
});
it('should match claude-opus45 (no separator after opus)', () => {
// The regex allows optional separators, so "45" can follow directly
// In practice, Anthropic uses separators, but regex is permissive
expect(reset('claude-opus45')).toBe(64000);
});
});
});
describe('maxOutputTokens.set()', () => {
const { set } = anthropicSettings.maxOutputTokens;
describe('Claude Sonnet and Haiku 4+ models (64K cap)', () => {
it('should cap at 64K for claude-sonnet-4 when value exceeds', () => {
expect(set(100000, 'claude-sonnet-4')).toBe(64000);
});
it('should allow 50K for claude-sonnet-4', () => {
expect(set(50000, 'claude-sonnet-4')).toBe(50000);
});
it('should cap at 64K for claude-haiku-4-5 when value exceeds', () => {
expect(set(80000, 'claude-haiku-4-5')).toBe(64000);
});
});
describe('Claude Opus 4.5+ models (64K cap)', () => {
it('should cap at 64K for claude-opus-4-5 when value exceeds', () => {
expect(set(100000, 'claude-opus-4-5')).toBe(64000);
});
it('should cap at model-specific 64K limit, not global 128K limit', () => {
// Values between 64K and 128K should be capped at 64K (model limit)
// This verifies the fix for the unreachable code issue
expect(set(70000, 'claude-opus-4-5')).toBe(64000);
expect(set(80000, 'claude-opus-4-5')).toBe(64000);
expect(set(100000, 'claude-opus-4-5')).toBe(64000);
expect(set(128000, 'claude-opus-4-5')).toBe(64000);
// Values above 128K should also be capped at 64K (not 128K)
expect(set(150000, 'claude-opus-4-5')).toBe(64000);
});
it('should allow 50K for claude-opus-4-5', () => {
expect(set(50000, 'claude-opus-4-5')).toBe(50000);
});
it('should cap at 64K for claude-opus-4-6', () => {
expect(set(80000, 'claude-opus-4-6')).toBe(64000);
});
it('should cap at 64K for claude-opus-5', () => {
expect(set(100000, 'claude-opus-5')).toBe(64000);
});
it('should cap at 64K for claude-opus-4-10', () => {
expect(set(100000, 'claude-opus-4-10')).toBe(64000);
});
});
describe('Claude Opus 4.0-4.4 models (32K cap)', () => {
it('should cap at 32K for claude-opus-4', () => {
expect(set(50000, 'claude-opus-4')).toBe(32000);
});
it('should allow 20K for claude-opus-4', () => {
expect(set(20000, 'claude-opus-4')).toBe(20000);
});
it('should cap at 32K for claude-opus-4-1', () => {
expect(set(50000, 'claude-opus-4-1')).toBe(32000);
});
it('should cap at 32K for claude-opus-4-4', () => {
expect(set(40000, 'claude-opus-4-4')).toBe(32000);
});
});
describe('Global 128K cap for all models', () => {
it('should cap at model-specific limit first, then global', () => {
// claude-sonnet-4 has 64K limit, so caps at 64K not 128K
expect(set(150000, 'claude-sonnet-4')).toBe(64000);
});
it('should cap at 128K for claude-3 models', () => {
expect(set(150000, 'claude-3-opus')).toBe(128000);
});
it('should cap at 128K for unknown models', () => {
expect(set(200000, 'unknown-model')).toBe(128000);
});
});
describe('Valid values within limits', () => {
it('should allow valid values for legacy models', () => {
expect(set(8000, 'claude-3-opus')).toBe(8000);
});
it('should allow 1 token minimum', () => {
expect(set(1, 'claude-opus-4-5')).toBe(1);
});
it('should allow 128K exactly', () => {
expect(set(128000, 'claude-3-opus')).toBe(128000);
});
});
});
});

View file

@ -386,6 +386,10 @@ export const anthropicSettings = {
return CLAUDE_4_64K_MAX_OUTPUT;
}
if (/claude-opus[-.]?(?:[5-9]|4[-.]?([5-9]|\d{2,}))/.test(modelName)) {
return CLAUDE_4_64K_MAX_OUTPUT;
}
if (/claude-opus[-.]?[4-9]/.test(modelName)) {
return CLAUDE_32K_MAX_OUTPUT;
}
@ -397,7 +401,14 @@ export const anthropicSettings = {
return CLAUDE_4_64K_MAX_OUTPUT;
}
if (/claude-(?:opus|haiku)[-.]?[4-9]/.test(modelName) && value > CLAUDE_32K_MAX_OUTPUT) {
if (/claude-opus[-.]?(?:[5-9]|4[-.]?([5-9]|\d{2,}))/.test(modelName)) {
if (value > CLAUDE_4_64K_MAX_OUTPUT) {
return CLAUDE_4_64K_MAX_OUTPUT;
}
return value;
}
if (/claude-opus[-.]?[4-9]/.test(modelName) && value > CLAUDE_32K_MAX_OUTPUT) {
return CLAUDE_32K_MAX_OUTPUT;
}