AI Model Context Window Capacities | Generated by AI
Here are the current maximum token capacities (context window sizes) for the models you’ve listed, based on recent documentation and announcements:
Anthropic Models
-
Claude Opus 4 / Opus 4.1 Offers a context window of 200,000 tokens, ideal for sustained reasoning and complex workflows (Anthropic, Amazon Web Services, Inc.).
-
Claude Sonnet 4 Also supports 200,000 tokens of context natively (Prompthub). Additionally, a beta upgrade now enables a massive 1 million tokens of context for enterprise/API users (Anthropic).
Mistral AI Models
-
Mistral Medium 3.1 Has a context window of 128,000 tokens (OpenRouter).
-
Mistral Large 2 Also supports a 128,000 token context length (Wikipedia).
-
Mistral Medium (earlier versions) Previously, the standard was 32,000 tokens for context (Clarifai, Mistral AI).
-
Mistral Small 3.1 Upgraded its context window from 32K to 128K tokens in this newer version (Wikipedia).
Additional Notes
-
A Wikipedia overview confirms that Claude 3 Opus had a 200K token context window, with expansions to 1 million tokens in certain cases (Wikipedia).
-
Mistral’s sliding window attention mechanism theoretically supports up to 4096 × 32 = 131,072 tokens, which closely aligns with the 128K figures reported (Hugging Face).
Summary (Concise Format)
- Claude Opus 4 / 4.1: 200,000 tokens
- Claude Sonnet 4: 200,000 tokens (standard), up to 1 million tokens (beta/enterprise)
- Mistral Medium 3.1: 128,000 tokens
- Mistral Large 2: 128,000 tokens
- Mistral Small 3.1: 128,000 tokens (upgraded from 32K)
Let me know if you’d like more details on other models or architecture-specific capabilities!