← API pricing · context windows
LLM context window comparison
How many tokens each of 39 live models takes in one request (prompt and reply together), the longest reply it can write, and where a maker starts charging more for a long prompt. Largest first. From each maker's own API listing in the models.dev catalog as of October 25, 2026, refreshed daily.
| Model | Context | Max output | Higher rate past | Input / output | AW Index |
|---|---|---|---|---|---|
| ChatGPT-5.6OpenAI | 1.05M | 128K | 272K | $4 / $20 | 50 |
| GPT-5.5OpenAI | 1.05M | 128K | 272K | $5 / $30 | 42.6 |
| GPT-6 AstraOpenAI | 1.05M | 128K | 272K | $10 / $50 | 87.4 |
| GPT-6 LunaOpenAI | 1.05M | 128K | 272K | $0.10 / $0.50 | 28 |
| GPT-6 SolOpenAI | 1.05M | 128K | 272K | $2 / $10 | 68.6 |
| GPT-6.1 SolOpenAI | 1.05M | 128K | 272K | $2 / $10 | 82.9 |
| DeepSeek V4 ProDeepSeek | 1.05M | 393K | — | $0.66 / $1.98 | 31.8 |
| Mistral Large 4Mistral | 1.05M | 262K | — | $1.36 / $4.18 | 32.6 |
| MiMo-V2.5Xiaomi | 1.05M | 131K | — | $0.14 / $0.28 | — |
| MiMo-V2.6 ProXiaomi | 1.05M | 131K | — | $0.435 / $0.87 | 65.1 |
| Muse Spark 1.1Meta | 1.05M | 131K | — | $1.25 / $4.25 | 30.8 |
| Muse Spark 1.2Meta | 1.05M | 131K | — | $1.25 / $4.25 | — |
| Muse Spark 1.3Meta | 1.05M | 131K | — | $1.25 / $4.25 | 74.9 |
| Gemini 3.5 FlashGoogle | 1.05M | 66K | — | $1.50 / $9 | 21.3 |
| Gemini 3.5 Flash-LiteGoogle | 1.05M | 66K | — | $0.30 / $2.50 | — |
| Gemini 3.6 FlashGoogle | 1.05M | 66K | — | $0.75 / $3.75 | — |
| Gemini 3.7 FlashGoogle | 1.05M | 66K | — | $0.75 / $3.75 | — |
| Gemini 3.8 FlashGoogle | 1.05M | 66K | — | $0.75 / $3.75 | 43.3 |
| Gemini 4 ArgonGoogle | 1.05M | 33K | — | $2 / $10 | 79.5 |
| Kimi K3Moonshot | 1.05M | — | — | $3 / $15 | 27.3 |
| Kimi K3.1 ProMoonshot | 1.05M | — | — | $2 / $10 | 86 |
| MiniMax M3MiniMax | 1M | 524K | 512K | $0.30 / $1.20 | 5.8 |
| MiniMax M3.1 FlashMiniMax | 1M | 524K | — | $0.15 / $0.60 | 48.8 |
| GLM-5.2Zhipu AI | 1M | 131K | — | $1.40 / $4.40 | 21.2 |
| GLM-5.3Zhipu AI | 1M | 131K | — | $1.40 / $4.40 | 60 |
| GLM-5.3 FlashZhipu AI | 1M | 131K | — | $0.15 / $0.50 | 32.4 |
| Qwen 3.8 FlashAlibaba | 1M | 131K | — | $0.15 / $0.47 | — |
| Qwen 3.8 MaxAlibaba | 1M | 131K | — | $2 / $6 | 62.9 |
| Claude Haiku 5.5Anthropic | 1M | 128K | 100K | $0.10 / $0.50 | 50.5 |
| Claude Opus 4.8Anthropic | 1M | 128K | — | $5 / $25 | 37.2 |
| Claude Opus 5Anthropic | 1M | 128K | — | $5 / $25 | 72.8 |
| Claude Opus 5.5Anthropic | 1M | 128K | — | $4 / $20 | 100 |
| Claude Sonnet 5Anthropic | 1M | 128K | — | $2 / $10 | 27.9 |
| Claude Sonnet 5.5Anthropic | 1M | 128K | — | $2 / $10 | 93.5 |
| Fable 5Anthropic | 1M | 128K | — | $10 / $50 | 72.1 |
| Fable 5.1Anthropic | 1M | 128K | — | $10 / $50 | 90.7 |
| Grok 4.5xAI | 500K | — | 200K | $2 / $6 | 34.9 |
| Grok 4.6xAI | 500K | — | 200K | $2 / $6 | 58.1 |
| Grok 4.7xAI | 500K | — | 200K | $2 / $6 | 59.4 |
Tokens; 1M is a million. Prices are per million tokens at the normal rate. “Higher rate past” is the prompt size at which the maker switches the whole request to its long-prompt price.
To compare only the models with room for a long document, filter the cost calculator to 1M-token windows.
Quick answers
Which LLM has the largest context window?
ChatGPT-5.6, GPT-5.5, GPT-6 Astra, GPT-6 Luna, GPT-6 Sol and GPT-6.1 Sol, at 1.05M tokens each, the largest of the 39 live models with a published first-party price.
How many LLMs have a 1 million token context window?
36 of the 39: ChatGPT-5.6, GPT-5.5, GPT-6 Astra, GPT-6 Luna and 32 more. The rest top out lower; the table lists each model's figure.
Which model can write the longest reply?
MiniMax M3 and MiniMax M3.1 Flash, with a maximum output of 524K tokens. A model's context window covers the prompt and the reply together, so the reply can never be longer than what the window leaves free.
Does a long prompt cost more?
On 11 of these models it does: past a threshold the maker bills a higher rate for the whole request. ChatGPT-5.6 past 272K ($8 input), GPT-5.5 past 272K ($10 input), GPT-6 Astra past 272K ($20 input), GPT-6 Luna past 272K ($0.20 input) and 7 more. The rest charge the same rate across their whole window.
Where do these numbers come from?
Each maker's own API listing in the models.dev catalog, as of October 25, 2026, for the model id that is the model's exact name. Some apps and plans cap the window lower than the API does.
Prices for every model: LLM API pricing. Scores: the bench index. For an alert when a new model goes live, watch free.
Cite this page
Writing about LLM context windows? Copy a link or a citation.
<a href="https://artificialwatch.com/llm-context-windows">LLM context window comparison</a>ArtificialWatch, "LLM context window comparison", https://artificialwatch.com/llm-context-windowsWindows keep growing. Get an alert when a new model lands.
Free: a Chrome push and an email about 15 minutes after a new model answers on a public API. Members get a text or a phone call the moment a launch is confirmed.
- Free forever
- No card
- Unsubscribe in one click