The quick answer
The largest context window in the current catalogue is 1M tokens, held by DeepSeek V4 Flash 0731 from DeepSeek. As a rough physical intuition, one token is about three quarters of an English word, and a manuscript page is about 400 words, so the table below translates every window into pages.
One row per major provider, showing its largest-context model:
| Provider | Largest-context model | Context window | Fits roughly |
|---|---|---|---|
| DeepSeek | DeepSeek V4 Flash 0731 | 1M tokens | about 2,500 pages |
| OpenAI | GPT-5.4 | 1M tokens | about 2,000 pages |
| Gemini 2.5 Flash | 1M tokens | about 2,000 pages | |
| Meta | Llama 4 Maverick | 1M tokens | about 2,000 pages |
| Qwen | Qwen3.8 2.4T A95B | 1M tokens | about 2,000 pages |
| Moonshot | Kimi K3 | 1M tokens | about 2,000 pages |
| Z.ai | GLM 5.2 | 1M tokens | about 2,000 pages |
| MiniMax | MiniMax M3 | 1M tokens | about 2,000 pages |
| Anthropic | Claude Fable 5 | 1M tokens | about 1,900 pages |
| xAI | Grok 4.3 | 1M tokens | about 1,900 pages |
| Mistral | Mistral Large 3 2512 | 262K tokens | about 492 pages |
If the question behind your search is "what is a context window in the first place", start with the plain-English explainer at What is a context window? and come back for the numbers.
Every model in the index (100 models)
The full catalogue, sorted by context window. The data comes from the same source as the AI Model Cost Index and was last fetched 2026-08-20; when the catalogue updates, this table updates with it. The credits column shows what each model costs per message inside Whizi, where a plan includes it.
| Model | Provider | Context window | Fits roughly | Credits in Whizi |
|---|---|---|---|---|
| DeepSeek V4 Flash 0731 | DeepSeek | 1M | about 2,500 pages | 1 |
| GPT-5.4 | OpenAI | 1M | about 2,000 pages | 15 |
| GPT-5.5 | OpenAI | 1M | about 2,000 pages | 20 |
| GPT-5.6 Luna | OpenAI | 1M | about 2,000 pages | 1 |
| GPT-5.6 Luna Pro | OpenAI | 1M | about 2,000 pages | Not metered |
| GPT-5.6 Sol | OpenAI | 1M | about 2,000 pages | 20 |
| GPT-5.6 Sol Pro | OpenAI | 1M | about 2,000 pages | Not metered |
| GPT-5.6 Terra | OpenAI | 1M | about 2,000 pages | 4 |
| GPT-5.6 Terra Pro | OpenAI | 1M | about 2,000 pages | Not metered |
| LongCat 2.0 | Meituan | 1M | about 2,000 pages | 1 |
| DeepSeek V4 Pro 0813 | DeepSeek | 1M | about 2,000 pages | 2 |
| Gemini 2.5 Flash | 1M | about 2,000 pages | 2 | |
| Gemini 2.5 Flash Lite | 1M | about 2,000 pages | 1 | |
| Gemini 3.1 Flash Lite | 1M | about 2,000 pages | 1 | |
| Gemini 3.1 Pro Preview | 1M | about 2,000 pages | 10 | |
| Gemini 3.5 Flash | 1M | about 2,000 pages | 8 | |
| Gemini 3.5 Flash Lite | 1M | about 2,000 pages | 2 | |
| Gemini 3.6 Flash | 1M | about 2,000 pages | 3 | |
| Gemini 3.7 Flash | 1M | about 2,000 pages | 2 | |
| GLM 5.2 | Z.ai | 1M | about 2,000 pages | 3 |
| GLM 5.3 | Z.ai | 1M | about 2,000 pages | 5 |
| Inkling | Thinkingmachines | 1M | about 2,000 pages | 4 |
| Kimi K3 | Moonshot | 1M | about 2,000 pages | 10 |
| Laguna S 2.1 | Poolside | 1M | about 2,000 pages | 1 |
| Llama 4 Maverick | Meta | 1M | about 2,000 pages | 1 |
| MiniMax M3 | MiniMax | 1M | about 2,000 pages | 1 |
| Muse Spark 1.1 | Meta | 1M | about 2,000 pages | 5 |
| Muse Spark 1.2 | Meta | 1M | about 2,000 pages | 5 |
| Qwen3.8 2.4T A95B | Qwen | 1M | about 2,000 pages | 8 |
| Claude Fable 5 | Anthropic | 1M | about 1,900 pages | 50 |
| Claude Opus 4.7 (Fast) | Anthropic | 1M | about 1,900 pages | Not metered |
| Claude Opus 4.8 | Anthropic | 1M | about 1,900 pages | 20 |
| Claude Opus 4.8 (Fast) | Anthropic | 1M | about 1,900 pages | Not metered |
| Claude Opus 5 | Anthropic | 1M | about 1,900 pages | 20 |
| Claude Opus 5 (Fast) | Anthropic | 1M | about 1,900 pages | Not metered |
| Claude Sonnet 4.5 | Anthropic | 1M | about 1,900 pages | 10 |
| Claude Sonnet 4.6 | Anthropic | 1M | about 1,900 pages | 10 |
| Claude Sonnet 5 | Anthropic | 1M | about 1,900 pages | 10 |
| Fugu Ultra | Sakana | 1M | about 1,900 pages | 25 |
| Grok 4.3 | xAI | 1M | about 1,900 pages | 4 |
| Qwen3.5 Plus 2026-04-20 | Qwen | 1M | about 1,900 pages | 1 |
| Qwen3.6 Flash | Qwen | 1M | about 1,900 pages | 1 |
| Qwen3.7 Flash | Qwen | 1M | about 1,900 pages | 1 |
| Qwen3.7 Max | Qwen | 1M | about 1,900 pages | 5 |
| Qwen3.7 Plus | Qwen | 1M | about 1,900 pages | 1 |
| Qwen3.8 27B | Qwen | 1M | about 1,900 pages | 3 |
| Qwen3.8 Max | Qwen | 1M | about 1,900 pages | 6 |
| Inkling Small | Thinkingmachines | 524K | about 983 pages | 1 |
| Solar Pro 4 | Upstage | 524K | about 983 pages | 1 |
| Nemotron 3 Ultra | NVIDIA | 512K | about 961 pages | 3 |
| Grok 4.5 | xAI | 500K | about 938 pages | 6 |
| Grok 4.6 | xAI | 500K | about 938 pages | 6 |
| GPT Chat Latest | OpenAI | 400K | about 750 pages | 25 |
| GPT-5.4 Mini | OpenAI | 400K | about 750 pages | 4 |
| GPT-5.4 Nano | OpenAI | 400K | about 750 pages | 2 |
| Nova Pro 1.0 | Amazon | 300K | about 563 pages | 3 |
| Hy3 | Tencent | 262K | about 492 pages | 1 |
| Kimi K2 Thinking | Moonshot | 262K | about 492 pages | 10 |
| Kimi K2.7 Code | Moonshot | 262K | about 492 pages | 10 |
| Laguna XS 2.1 | Poolside | 262K | about 492 pages | 1 |
| Ling-3.0-flash | inclusionAI | 262K | about 492 pages | 1 |
| Mistral Large 3 2512 | Mistral | 262K | about 492 pages | 2 |
| Mistral Medium 3.5 | Mistral | 262K | about 492 pages | 6 |
| Nemotron 3.5 Lightning | NVIDIA | 262K | about 492 pages | 1 |
| Nex-N2-Mini | Nex Agi | 262K | about 492 pages | 1 |
| Nex-N2-Pro | Nex Agi | 262K | about 492 pages | 1 |
| Qwen3 Coder Next | Qwen | 262K | about 492 pages | 1 |
| Qwen3.6 35B A3B | Qwen | 262K | about 492 pages | 1 |
| Ring-2.6-1T | inclusionAI | 262K | about 492 pages | 1 |
| Sakana Namazu | Sakana | 262K | about 492 pages | 4 |
| Seed 2.1 Turbo | Bytedance Seed | 262K | about 492 pages | 2 |
| Seed-2.0-Code | Bytedance Seed | 262K | about 492 pages | 3 |
| Step 3.7 Flash | Stepfun | 262K | about 492 pages | 1 |
| Codestral 2508 | Mistral | 256K | about 480 pages | 1 |
| Command A | Cohere | 256K | about 480 pages | 10 |
| Grok Build 0.1 | xAI | 256K | about 480 pages | 3 |
| KAT-Coder-Air V2.5 | Kwaipilot | 256K | about 480 pages | 1 |
| KAT-Coder-Pro V2.5 | Kwaipilot | 256K | about 480 pages | 3 |
| GLM 4.6 | Z.ai | 205K | about 384 pages | 1 |
| GLM 4.7 | Z.ai | 205K | about 384 pages | 2 |
| GLM 5 | Z.ai | 205K | about 384 pages | 2 |
| MiniMax M2 | MiniMax | 205K | about 384 pages | 1 |
| Claude Haiku 4.5 | Anthropic | 200K | about 375 pages | 4 |
| DeepSeek V3.1 | DeepSeek | 164K | about 307 pages | 1 |
| DeepSeek V3.2 | DeepSeek | 164K | about 307 pages | 1 |
| Aion-3.0 | Aion Labs | 131K | about 246 pages | 10 |
| Aion-3.0-Mini | Aion Labs | 131K | about 246 pages | 2 |
| Granite 4.1 8B | Ibm Granite | 131K | about 246 pages | 1 |
| Kimi K2 0711 | Moonshot | 131K | about 246 pages | 10 |
| Llama 3.3 70B Instruct | Meta | 131K | about 246 pages | 1 |
| Mistral Medium 3.1 | Mistral | 131K | about 246 pages | 2 |
| Muse Glimmer 30B | Meta | 131K | about 246 pages | 2 |
| Nano Banana 2 (Gemini 3.1 Flash Image) | 131K | about 246 pages | Not metered | |
| Nano Banana Pro (Gemini 3 Pro Image) | 131K | about 246 pages | Not metered | |
| Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) | 66K | about 123 pages | Not metered | |
| R1 | DeepSeek | 64K | about 120 pages | 2 |
| Perceptron Mk1 | Perceptron | 33K | about 61 pages | 1 |
| Phi 4 | Microsoft | 16K | about 31 pages | 1 |
| Hy-MT2-1.8B | Tencent | 8K | about 15 pages | Not metered |
| Hy-MT2-30B-A3B | Tencent | 8K | about 15 pages | Not metered |
How to read these numbers honestly
The window is shared. Everything counts against it at once: your prompt, every earlier message in the conversation, any documents you attached, and the answer being written. A model with a 128K window that has already read a 100K-token contract does not have 128K left for the discussion; it has 28K.
Advertised is not effective. Long-context research keeps finding the same pattern: models recall the start and end of a huge context better than the middle. A model can accept a million tokens and still gloss over the clause buried at token 400,000. For high-stakes work on long documents, ask specifically about the middle sections, or feed the document in labelled parts.
Bigger is not automatically better. Large windows cost more to run and can dilute attention across irrelevant text. If your work is emails and short drafts, the window size will never be the thing you notice. It starts to matter at whole contracts, codebases, transcripts, and books.
Tokens are not words. The three-quarters rule is a good English average, but code, numbers, and non-English languages often tokenize less efficiently, so the page estimates in the table are deliberately rough.
Using the right window for the job
The practical conclusion from the table is that context window is a per-task choice, not a per-subscription choice. The model with the biggest window is rarely the best writer, and the best writer is rarely the cheapest way to summarize 800 pages.
That is the argument for having the catalogue in one workspace. In Whizi you can hand the 200-page PDF to a large-context model, then switch to Claude or GPT in the same conversation to draft from what it found, without the document leaving the thread. The how to choose a model guide covers the same decision for quality and cost.
For citing this page: the anchor-free URL is stable, the data date is printed above the full table, and the token figures come from the provider catalogue rather than from marketing pages.
- Subtract your documents and history from the window before trusting the headline number
- For long documents, test recall from the middle, not just the start
- Translate tokens to pages with the three-quarters rule before deciding anything
- Pick the window per task instead of per subscription
- Check the data date above the full table when citing a figure
Frequently asked questions
Which AI model has the largest context window?
In the current catalogue it is DeepSeek V4 Flash 0731 from DeepSeek at 1M tokens. The full table above lists all 100 models sorted by window size, with the data date printed above it.
What is the largest Claude context window?
The largest Claude context window in the index is 1M tokens (Claude Fable 5), which fits about 1,900 pages of text. Smaller Claude variants are listed in the full table above.
What is the largest GPT context window?
The largest GPT context window in the index is 1M tokens (GPT-5.4), which fits about 2,000 pages of text. Smaller GPT variants are listed in the full table above.
What is the largest Gemini context window?
The largest Gemini context window in the index is 1M tokens (Gemini 2.5 Flash), which fits about 2,000 pages of text. Smaller Gemini variants are listed in the full table above.
What is the largest Grok context window?
The largest Grok context window in the index is 1M tokens (Grok 4.3), which fits about 1,900 pages of text. Smaller Grok variants are listed in the full table above.
Is a bigger context window always better?
No. Larger windows cost more, and recall degrades toward the middle of a very large context. Bigger is decisive for whole books, codebases, and long transcripts, and irrelevant for everyday chat. Match the window to the task rather than paying for the biggest number.
How many pages fit in a context window?
Divide the token count by roughly 530 to get manuscript pages: one token is about three quarters of an English word, and a page is about 400 words. So 128K tokens is about 240 pages, and a million tokens is about 1,900 pages. Code and non-English text usually fit less.