LLM Context Window Comparison

See how much text each model can take in one request. The chart compares context windows (or max output) across OpenAI, Anthropic, Google, Mistral, Meta, DeepSeek and xAI, and the fit check shows which models can hold your prompt.

How to compare context windows

1. Pick providers, and switch between context window and maximum output tokens.
2. Type your prompt size (e.g. 150k) to see which models can hold it — models that can't turn red.
3. Check the full table below for exact limits, including separate maximum-input caps.

What a context window is

The context window is the maximum number of tokens a model can consider in one request: your system prompt, the conversation history, any documents or tool results, and the response it generates. If input plus output exceeds it, the request fails or old messages must be dropped. Some models also cap input separately — GPT-5 has a 400K window but accepts at most 272K input tokens.

All context windows

Bigger isn't automatically better

• Cost scales with what you send. A full 1M-token prompt on GPT-6 Sol costs $4.00 per call — and long-context surcharges kick in above 272K on OpenAI and 200K on Gemini Pro and Grok.
• Recall drops in the middle. Models retrieve facts near the start and end of long prompts more reliably; put key instructions last.
• Latency grows with prompt length. Retrieval (RAG) that sends only relevant chunks is often faster and cheaper than stuffing everything in.

Context sizes in tokens: 1M tokens ≈ 750,000 English words ≈ 1,500 pages. Count your own text with the token counter.

Related tools

• LLM Token Counter
• LLM API Pricing Calculator
• Claude Token Counter
• GPT Token Counter