Context Window Comparison
48 LLMs ranked by context window size. Larger context means more text can be processed in a single conversation — crucial for analysing long documents, codebases, and research papers.
11
1M+ tokens
16
200K–999K
20
32K–199K
1
<32K tokens
Largest Context by Tier
Context Size Ranking
Full Table
| # | Model | Context | Quality evidence |
|---|---|---|---|
| 1 | Llama 4 Scout OSS Meta | 1.3M | Not published |
| 2 | Gemini 2.5 Pro Google | 1.0M | Not published |
| 3 | Gemini 2.5 Flash (batch) Google | 1.0M | Not published |
| 4 | Llama 4 Maverick OSS Meta | 1.0M | Not published |
| 5 | Gemini 2.5 Flash Lite (batch) Google | 1.0M | Not published |
| 6 | GPT-4.1 (batch) OpenAI | 1.0M | Not published |
| 7 | GPT-4.1 Mini (batch) OpenAI | 1.0M | Not published |
| 8 | GPT-4.1 Nano (batch) OpenAI | 1.0M | Not published |
| 9 | Claude Opus 4.6 (batch) Anthropic | 1M | Not published |
| 10 | Claude Sonnet 4.6 (batch) Anthropic | 1M | Not published |
| 11 | Claude Sonnet 4.5 (batch) Anthropic | 1M | Not published |
| 12 | GPT-5.2 Pro (batch) OpenAI | 400K | Not published |
| 13 | GPT-5.2 OpenAI | 400K | Not published |
| 14 | GPT-5 Pro (batch) OpenAI | 400K | Not published |
| 15 | GPT-5 (batch) OpenAI | 400K | Not published |
| 16 | GPT-5 Nano (batch) OpenAI | 400K | Not published |
| 17 | Nova Pro 1.0 Amazon | 300K | Not published |
| 18 | Nova Lite 1.0 Amazon | 300K | Not published |
| 19 | Command A OSS Cohere | 256K | Not published |
| 20 | O4 Mini OpenAI | 200K | Not published |
| 21 | Claude Opus 4.5 (batch) Anthropic | 200K | Not published |
| 22 | O3 OpenAI | 200K | Not published |
| 23 | O3 Pro OpenAI | 200K | Not published |
| 24 | o1 OpenAI | 200K | Not published |
| 25 | o3 Mini (batch) OpenAI | 200K | Not published |
| 26 | Claude Sonnet 4 Anthropic | 200K | Not published |
| 27 | Claude Haiku 4.5 (batch) Anthropic | 200K | Not published |
| 28 | R1 0528 OSS DeepSeek | 163.8K | Not published |
| 29 | DeepSeek V3.2 OSS DeepSeek | 163.8K | Not published |
| 30 | DeepSeek V3 OSS DeepSeek | 163.8K | Not published |
| 31 | Llama 3.1 70B Instruct OSS Meta | 131.1K | Not published |
| 32 | Mistral Nemo OSS Mistral | 131.1K | Not published |
| 33 | Llama 3.3 70B Instruct OSS Meta | 131.1K | Not published |
| 34 | Llama 3.1 8B Instruct OSS Meta | 131.1K | Not published |
| 35 | Command R+ (08-2024) OSS Cohere | 128K | Not published |
| 36 | Mistral Small 3.1 24B OSS Mistral | 128K | Not published |
| 37 | GPT-4o (batch) OpenAI | 128K | Not published |
| 38 | GPT-4o-mini (batch) OpenAI | 128K | Not published |
| 39 | Command R (08-2024) OSS Cohere | 128K | Not published |
| 40 | Mistral Large OSS Mistral | 128K | Not published |
| 41 | Nova Micro 1.0 Amazon | 128K | Not published |
| 42 | Command R7B (12-2024) OSS Cohere | 128K | Not published |
| 43 | Sonar Perplexity | 127.1K | Not published |
| 44 | Reka Flash 3 Reka | 65.5K | Not published |
| 45 | DeepSeek R1 OSS DeepSeek | 64K | Not published |
| 46 | Qwen2.5 Coder 32B Instruct OSS Alibaba | 32.8K | Not published |
| 47 | Qwen2.5 72B Instruct OSS Alibaba | 32.8K | Not published |
| 48 | Phi 4 OSS Microsoft | 16.4K | Not published |
What is a context window?
The context window is the maximum amount of text a model can process in a single request — including both your input and its response. It's measured in tokens (roughly 0.75 words per token).
32K tokens
~24,000 words / ~96 pages. Enough for a long article or short story.
200K tokens
~150,000 words / ~600 pages. Enough for a full novel or large codebase.
1M+ tokens
~750,000+ words / ~3,000+ pages. Entire book series, repos, or document collections.