Context Window Visualizer
See how much each model can process in a single request.
Compare context window sizes across every major LLM in one interactive visualization. See which models can handle entire codebases, novels, or long document collections. Covering GPT-5.4 (1.05M tokens), Gemini 3.1 Pro (1M tokens), Claude Opus 4.6 (200K tokens), and more.
Context Window Sizes by Model
What Fits in Each Context Window?
| Document Type | Tokens | Words |
|---|---|---|
| A tweet | 70 | ~53 |
| A short email | 375 | ~281 |
| A 5-page PDF | 3.5K | ~2,625 |
| A blog post (1,000 words) | 1.5K | ~1,125 |
| A 50-page report | 35K | ~26,250 |
| A full codebase (10K LOC) | 114K | ~85,500 |
| A novel (80K words) | 120K | ~90,000 |
All Model Context Windows
| Model | Provider | Context Window | ≈ Words | Max Output |
|---|---|---|---|---|
| Grok 4.1 Fast | xAI | 2M | ~1,500,000 | 16K |
| GPT-5.4 | OpenAI | 1.1M | ~787,500 | 128K |
| GPT-5.4 Pro | OpenAI | 1.1M | ~787,500 | 128K |
| Llama 4 Maverick | Meta | 1.0M | ~786,432 | 65.5K |
| GPT-4.1 | OpenAI | 1.0M | ~785,682 | 32.8K |
| GPT-4.1 mini | OpenAI | 1.0M | ~785,682 | 32.8K |
| GPT-4.1 nano | OpenAI | 1.0M | ~785,682 | 32.8K |
| Gemini 3.1 Pro | 1M | ~750,000 | 65.5K | |
| Gemini 3.1 Flash Lite | 1M | ~750,000 | 65.5K | |
| Gemini 3 Flash | 1M | ~750,000 | 65.5K | |
| Gemini 2.5 Pro | 1M | ~750,000 | 65.5K | |
| Gemini 2.5 Flash | 1M | ~750,000 | 65.5K | |
| Gemini 2.5 Flash Lite | 1M | ~750,000 | 65.5K | |
| Nova Premier | Amazon | 1M | ~750,000 | 32.8K |
| Nova 2 Lite | Amazon | 1M | ~750,000 | 32.8K |
| GPT-5.3 Codex | OpenAI | 400K | ~300,000 | 128K |
| GPT-5.2 | OpenAI | 400K | ~300,000 | 128K |
| GPT-5.2 Pro | OpenAI | 400K | ~300,000 | 128K |
| GPT-5.1 | OpenAI | 400K | ~300,000 | 128K |
| GPT-5 | OpenAI | 400K | ~300,000 | 128K |
| GPT-5 Pro | OpenAI | 400K | ~300,000 | 128K |
| GPT-5 mini | OpenAI | 400K | ~300,000 | 128K |
| GPT-5 nano | OpenAI | 400K | ~300,000 | 128K |
| Llama 4 Scout | Meta | 327.7K | ~245,760 | 32.8K |
| Nova Pro | Amazon | 300K | ~225,000 | 5.1K |
| Nova Lite | Amazon | 300K | ~225,000 | 5.1K |
| Mistral Large 3 | Mistral | 262.1K | ~196,608 | 32.8K |
| Devstral 2 | Mistral | 262.1K | ~196,608 | 32.8K |
| Grok 4 | xAI | 256K | ~192,000 | 32.8K |
| Grok Code Fast | xAI | 256K | ~192,000 | 32.8K |
| Codestral | Mistral | 256K | ~192,000 | 32.8K |
| Command A | Cohere | 256K | ~192,000 | 8.2K |
| o3 | OpenAI | 200K | ~150,000 | 100K |
| o3 Pro | OpenAI | 200K | ~150,000 | 100K |
| o4 mini | OpenAI | 200K | ~150,000 | 100K |
| o3 mini | OpenAI | 200K | ~150,000 | 100K |
| o1 | OpenAI | 200K | ~150,000 | 100K |
| Claude Opus 4.6 | Anthropic | 200K | ~150,000 | 128K |
| Claude Sonnet 4.6 | Anthropic | 200K | ~150,000 | 64K |
| Claude Opus 4.5 | Anthropic | 200K | ~150,000 | 64K |
| Claude Sonnet 4.5 | Anthropic | 200K | ~150,000 | 64K |
| Claude Haiku 4.5 | Anthropic | 200K | ~150,000 | 64K |
| Claude Sonnet 4 | Anthropic | 200K | ~150,000 | 64K |
| Claude Haiku 3.5 | Anthropic | 200K | ~150,000 | 8.2K |
| Grok 3 | xAI | 131.1K | ~98,304 | 16.4K |
| Grok 3 mini | xAI | 131.1K | ~98,304 | 16.4K |
| Mistral Medium 3.1 | Mistral | 131.1K | ~98,304 | 32.8K |
| Pixtral Large | Mistral | 131.1K | ~98,304 | 32.8K |
| Mistral Small 3.2 | Mistral | 131.1K | ~98,304 | 32.8K |
| Mistral Nemo | Mistral | 131.1K | ~98,304 | 32.8K |
| GPT-4o | OpenAI | 128K | ~96,000 | 16.4K |
| GPT-4o mini | OpenAI | 128K | ~96,000 | 16.4K |
| DeepSeek V3.2 | DeepSeek | 128K | ~96,000 | 8.2K |
| DeepSeek V3.2 Reasoner | DeepSeek | 128K | ~96,000 | 64K |
| Nova Micro | Amazon | 128K | ~96,000 | 5.1K |
| Command R | Cohere | 128K | ~96,000 | 4.1K |
| Command R7B | Cohere | 128K | ~96,000 | 4.1K |
| Magistral Medium | Mistral | 40K | ~30,000 | 32.8K |
Why Context Window Size Matters
A model's context window determines how much text it can process in a single request — including your prompt, any documents you provide, and the model's response. Larger context windows enable processing entire codebases, long legal documents, or multi-chapter reports without splitting them into chunks. However, larger contexts typically cost more and may increase latency. GPT-5.4 leads with 1,050,000 tokens, followed by Gemini 3.1 Pro at 1,000,000 and Claude Opus 4.6 at 200,000.
What Fits in Each Context Size
A 128K context window (GPT-4o, DeepSeek-R1) holds approximately 96,000 words — equivalent to a full novel. A 200K context window (Claude models) holds about 150,000 words or a thick technical manual. A 1M+ context window (GPT-5.4, Gemini 3.1 Pro) holds roughly 750,000 words — enough for multiple textbooks or an entire medium-sized codebase. Understanding these capacities helps you choose the right model for your document processing needs.
Frequently Asked Questions
Which LLM has the largest context window in 2026?
Can I fit an entire codebase in an LLM context window?
Does a larger context window cost more?
What is a context window in LLM terms?
Context window sizes sourced from official provider documentation. Effective usable context may be slightly less than the maximum due to system overhead.