llm-info

Information on LLM models, context window token limit, output token limit, pricing and more, developed by 16x Prompt and 16x Eval team.

Related projects:

• send-prompt: A unified TypeScript library for AI model interactions with standardized interfaces and function calling.
• ai-file-edit: A library for editing files using AI models such as GPT, Claude, and Gemini.

Information provided

• context window token limit
• output token limit
• pricing
• image input support
• tokenizerId for loading tokenizer from @xenova/transformers
• legacy status and legacy reason
• alpha/beta status
• small model indicator
• reasoning model indicator
• coding recommendation
• writing recommendation
• OpenRouter model ID mapping
• notes and documentation URLs
• API keys page URLs for easy access to provider API key management
• Deprecated model mappings (DeepSeek-V3 under deepseek-chat, DeepSeek-R1 under deepseek-reasoner)
• and more

Models included

• GPT-5.2
• GPT-5.2 Pro
• o3
• o4-mini
• Claude Opus 4.5
• Claude Sonnet 4.5
• Claude Haiku 4.5
• DeepSeek-V3.1
• Gemini 3 Pro Preview
• Gemini 2.5 Flash
• Grok 4
• .. and more

Non-models (model-like) included:

• ChatGPT

Providers supported

First-party providers:

• OpenAI
• Anthropic
• Azure OpenAI
• DeepSeek
• Google
• xAI

Third-party providers:

• OpenRouter
• Fireworks

Install:

$ npm install llm-info

Usage

// Models
import { AllModels, ModelEnum, NonModelEnum, ModelInfoMap } from 'llm-info';

console.log(AllModels);
/*
[
  'gpt-4',
  'gpt-4-turbo',
  'gpt-4o',
  'gpt-4o-64k-output-alpha',
  'gpt-4o-mini',
  'gpt-4o-2024-08-06',
  'o1-preview',
  'o1-mini',
  'o1',
  'claude-3-5-sonnet-20240620',
  'claude-3-5-sonnet-20241022',
  'claude-3-5-haiku-20241022',
  // ...
]
*/

// Model Info
const modelInfo = ModelInfoMap['gpt-4o'];
console.log(modelInfo);
/*
{
  name: 'GPT-4o',
  provider: 'openai',
  contextWindowTokenLimit: 128000,
  outputTokenLimit: 4096,
  pricePerMillionInputTokens: 5,
  pricePerMillionOutputTokens: 15,
  tokenizerId: 'Xenova/gpt-4o',
  supportsImageInput: true,
  legacy: false
}
*/

// Example with more fields including recommendations
const recentModelInfo = ModelInfoMap['claude-sonnet-4-20250514'];
console.log(recentModelInfo);
/*
{
  name: 'Claude Sonnet 4',
  provider: 'anthropic',
  contextWindowTokenLimit: 200000,
  outputTokenLimit: 64000,
  pricePerMillionInputTokens: 3,
  pricePerMillionOutputTokens: 15,
  tokenizerId: 'Xenova/claude-tokenizer',
  notes: 'Optimal balance of intelligence, cost, and speed',
  notesUrl: 'https://www.anthropic.com/claude/sonnet',
  supportsImageInput: true,
  recommendedForCoding: true,
  recommendedForWriting: true,
  legacy: false,
  openRouterModelId: 'anthropic/claude-sonnet-4'
}
*/

// Providers
import { AI_PROVIDER_NAME_MAP } from 'llm-info';

console.log(AI_PROVIDER_NAME_MAP);
/*
{
  openai: 'OpenAI',
  anthropic: 'Anthropic',
  'azure-openai': 'Azure OpenAI',
  deepseek: 'DeepSeek',
  openrouter: 'OpenRouter',
  google: 'Google',
  fireworks: 'Fireworks'
}
*/

// Provider Configuration
import { AI_PROVIDER_CONFIG, AI_PROVIDERS } from 'llm-info';

console.log(AI_PROVIDER_CONFIG);
/*
{
  openai: {
    name: 'OpenAI',
    firstParty: true,
    apiKeysPage: 'https://platform.openai.com/api-keys'
  },
  anthropic: {
    name: 'Anthropic',
    firstParty: true,
    apiKeysPage: 'https://console.anthropic.com/settings/keys'
  },
  'azure-openai': {
    name: 'Azure OpenAI',
    firstParty: false,
    baseURL: 'https://<your-resource-name>.openai.azure.com/',
    website: 'https://azure.microsoft.com/en-us/products/ai-services/openai-service'
  },
  deepseek: {
    name: 'DeepSeek',
    firstParty: true,
    baseURL: 'https://api.deepseek.com',
    apiKeysPage: 'https://platform.deepseek.com/api_keys'
  },
  openrouter: {
    name: 'OpenRouter',
    firstParty: false,
    baseURL: 'https://openrouter.ai/api/v1',
    website: 'https://openrouter.ai',
    apiKeysPage: 'https://openrouter.ai/keys'
  },
  google: {
    name: 'Google',
    firstParty: true,
    baseURL: 'https://generativelanguage.googleapis.com/v1beta/openai/',
    apiKeysPage: 'https://aistudio.google.com/app/apikey'
  },
  fireworks: {
    name: 'Fireworks',
    firstParty: false,
    baseURL: 'https://api.fireworks.ai/inference/v1',
    website: 'https://fireworks.ai'
  },
  xai: {
    name: 'xAI',
    firstParty: true,
    baseURL: 'https://api.x.ai/v1',
    apiKeysPage: 'https://console.x.ai/'
  }
}
*/

// Tokenizer
import { AutoTokenizer } from '@xenova/transformers';
const testSentence =
  "Many words map to one token, but some don't: indivisible.";
const results: string[] = [];
for (let i = 0; i < AllModels.length; i++) {
  const model = AllModels[i];
  if (ModelInfoMap[model].tokenizerId) {
    const tokenizer = await AutoTokenizer.from_pretrained(
      ModelInfoMap[model].tokenizerId
    );
    const tokens = tokenizer.encode(testSentence);
    results.push(`${model}: ${tokens.length}`);
  }
}
console.log(`Test sentence: ${testSentence}\n${results.join('\n')}`);
// Test sentence: Many words map to one token, but some don't: indivisible.
// gpt-4: 15
// gpt-4o: 14
// gpt-4o-mini: 14
// claude-3-5-sonnet-20240620: 16

Deprecated Model Mappings

Some model IDs have been reused for newer versions. The library includes a deprecated mapping system to track these transitions:

import {
  getDeprecatedMappingByModelId,
  getAllDeprecatedMappings,
} from 'llm-info';

// Check for deprecated model mappings
const deprecatedMapping = getDeprecatedMappingByModelId('deepseek-chat');
if (deprecatedMapping) {
  console.log(`Model ID: ${deprecatedMapping.modelId}`);
  console.log(`Deprecated Name: ${deprecatedMapping.deprecatedName}`);
  console.log(`Deprecated Release: ${deprecatedMapping.deprecatedReleaseDate}`);
  console.log(`Transition Date: ${deprecatedMapping.transitionDate}`);
  console.log('Deprecated Model Info:', deprecatedMapping.deprecatedModelInfo);
}
/*
Model ID: deepseek-chat
Deprecated Name: DeepSeek-V3 (new)
Deprecated Release: 2025-03-24
Transition Date: 2025-08-21
Deprecated Model Info: {
  id: 'deepseek-chat',
  name: 'DeepSeek-V3 (new)',
  provider: 'deepseek',
  contextWindowTokenLimit: 64000,
  outputTokenLimit: 8192,
  pricePerMillionInputTokens: 0.27,
  pricePerMillionOutputTokens: 1.1,
  tokenizerId: null,
  legacy: true,
  legacyReason: 'Replaced by DeepSeek-V3.1 on the same model ID',
  releaseDate: '2025-03-24'
}
*/

// Get all deprecated models info
const allModelsInfo = getAllDeprecatedModelsInfo();
console.log(allModelsInfo);
/*
[
    {
      id: 'deepseek-chat',
      name: 'DeepSeek-V3 (new)',
      provider: 'deepseek',
      contextWindowTokenLimit: 64000,
      outputTokenLimit: 8192,
      pricePerMillionInputTokens: 0.27,
      pricePerMillionOutputTokens: 1.1,
      tokenizerId: null,
      legacy: true,
      legacyReason: 'Replaced by DeepSeek-V3.1 on the same model ID',
      releaseDate: '2025-03-24'
    },
    {
      id: 'deepseek-reasoner',
      name: 'DeepSeek-R1',
      provider: 'deepseek',
      contextWindowTokenLimit: 64000,
      outputTokenLimit: 8000,
      pricePerMillionInputTokens: 0.55,
      pricePerMillionOutputTokens: 2.19,
      tokenizerId: 'Xenova/gpt-4o',
      reasoning: true,
      supportsImageInput: false,
      legacy: true,
      legacyReason: 'Replaced by DeepSeek-V3.1 (Thinking Mode) on the same model ID',
      releaseDate: '2025-05-28'
    }
  ]
*/

Current Deprecated Mappings

• deepseek-chat: Originally "DeepSeek-V3 (new)" (released 2025-03-24), now refers to "DeepSeek-V3.1" (transitioned 2025-08-21)
• deepseek-reasoner: Originally "DeepSeek-R1" (released 2025-05-28), now refers to "DeepSeek-V3.1 (Thinking Mode)" (transitioned 2025-08-21)

Testing

$ npm test