The biggest context window on CoreAI right now belongs to xAI: Grok 4.20 Multi-Agent at 2000K tokens — enough to hold several full-length books in a single conversation. 50 models now offer a context window of one million tokens or more. A model's context window is its working memory: everything you paste, attach, and discuss must fit inside it, so a bigger window means the model can reason over entire codebases, contracts, research papers, or transcripts without losing track.
As a rule of thumb, 1,000 tokens is about 750 words. A 128K-token model holds roughly a 350-page book; a 1M-token model holds around eight of them. The ranking below comes straight from the live catalog.
| Model | Provider | Context | Input $/1M | Output $/1M | Capabilities |
|---|---|---|---|---|---|
| xAI: Grok 4.20 Multi-Agent | xAI | 2000K | $1.25 | $2.50 | Reasoning, Vision |
| xAI: Grok 4.20 | xAI | 2000K | $1.25 | $2.50 | Reasoning, Vision |
| OpenAI: GPT-5.6 Luna Pro | OpenAI | 1050K | $1 | $6 | Reasoning, Vision |
| OpenAI: GPT-5.6 Luna | OpenAI | 1050K | $1 | $6 | Reasoning, Vision |
| OpenAI: GPT-5.6 Terra Pro | OpenAI | 1050K | $2.50 | $15 | Reasoning, Vision |
| OpenAI: GPT-5.6 Terra | OpenAI | 1050K | $2.50 | $15 | Reasoning, Vision |
| OpenAI: GPT-5.6 Sol Pro | OpenAI | 1050K | $5 | $30 | Reasoning, Vision |
| OpenAI: GPT-5.6 Sol | OpenAI | 1050K | $5 | $30 | Reasoning, Vision |
| OpenAI GPT Latest | ~openai | 1050K | $5 | $30 | Reasoning, Vision |
| OpenAI: GPT-5.5 Pro | OpenAI | 1050K | $30 | $180 | Reasoning, Vision |
| OpenAI: GPT-5.5 | OpenAI | 1050K | $5 | $30 | Reasoning, Vision |
| OpenAI: GPT-5.4 Pro | OpenAI | 1050K | $30 | $180 | Reasoning, Vision |
| OpenAI: GPT-5.4 | OpenAI | 1050K | $2.50 | $15 | Reasoning, Vision |
| Z.ai: GLM 5.2 | Z-ai | 1049K | $0.86 | $2.69 | Reasoning |
| Google: Gemini 3.5 Flash | 1049K | $1.50 | $9 | Reasoning, Vision | |
| Google: Gemini 3.1 Flash Lite | 1049K | $0.25 | $1.50 | Reasoning, Vision | |
| Google Gemini Pro Latest | 1049K | $2 | $12 | Reasoning, Vision | |
| Google Gemini Flash Latest | 1049K | $1.50 | $9 | Reasoning, Vision | |
| DeepSeek: DeepSeek V4 Pro | DeepSeek | 1049K | $0.43 | $0.87 | Reasoning |
| Xiaomi: MiMo-V2.5-Pro | Xiaomi | 1049K | $0.43 | $0.87 | Reasoning |
| Xiaomi: MiMo-V2.5 | Xiaomi | 1049K | $0.14 | $0.28 | Reasoning, Vision |
| Google: Lyria 3 Pro Preview | 1049K | $0 | $0 | Vision | |
| Google: Lyria 3 Clip Preview | 1049K | $0 | $0 | Vision | |
| Google: Gemini 3.1 Flash Lite Preview | 1049K | $0.25 | $1.50 | Reasoning, Vision | |
| Google: Gemini 3.1 Pro Preview Custom Tools | 1049K | $2 | $12 | Reasoning, Vision | |
| Google: Gemini 3.1 Pro Preview | 1049K | $2 | $12 | Reasoning, Vision | |
| Google: Gemini 3 Flash Preview | 1049K | $0.50 | $3 | Reasoning, Vision | |
| Google: Gemini 2.5 Flash Lite | 1049K | $0.10 | $0.40 | Reasoning, Vision | |
| Google: Gemini 2.5 Flash | 1049K | $0.30 | $2.50 | Reasoning, Vision | |
| Google: Gemini 2.5 Pro | 1049K | $1.25 | $10 | Reasoning, Vision | |
| Google: Gemini 2.5 Pro Preview 06-05 | 1049K | $1.25 | $10 | Reasoning, Vision | |
| Google: Gemini 2.5 Pro Preview 05-06 | 1049K | $1.25 | $10 | Reasoning, Vision | |
| Meta: Llama 4 Maverick | Meta | 1049K | $0.20 | $0.80 | Vision |
| DeepSeek: DeepSeek V4 Flash | DeepSeek | 1049K | $0.10 | $0.20 | Reasoning |
| OpenAI: GPT-4.1 | OpenAI | 1048K | $2 | $8 | Vision |
| OpenAI: GPT-4.1 Mini | OpenAI | 1048K | $0.40 | $1.60 | Vision |
| OpenAI: GPT-4.1 Nano | OpenAI | 1048K | $0.10 | $0.40 | Vision |
| Writer: Palmyra X5 | Writer | 1040K | $0.60 | $6 | — |
| MiniMax: MiniMax-01 | Minimax | 1000K | $0.20 | $1.10 | Vision |
| Anthropic: Claude Sonnet 5 | Anthropic | 1000K | $2 | $10 | Reasoning, Vision |
| Sakana: Fugu Ultra | Sakana | 1000K | $5 | $30 | Reasoning, Web search, Vision |
| Anthropic: Claude Fable Latest | ~anthropic | 1000K | $10 | $50 | Reasoning, Vision |
| Anthropic: Claude Fable 5 | Anthropic | 1000K | $10 | $50 | Reasoning, Vision |
| NVIDIA: Nemotron 3 Ultra (free) | NVIDIA | 1000K | $0 | $0 | Reasoning |
| Qwen: Qwen3.7 Plus | Qwen | 1000K | $0.32 | $1.28 | Reasoning, Vision |
| Anthropic: Claude Opus 4.8 (Fast) | Anthropic | 1000K | $10 | $50 | Reasoning, Vision |
| Anthropic: Claude Opus 4.8 | Anthropic | 1000K | $5 | $25 | Reasoning, Vision |
| Qwen: Qwen3.7 Max | Qwen | 1000K | $1.25 | $3.75 | Reasoning |
| Anthropic: Claude Opus 4.7 (Fast) | Anthropic | 1000K | $30 | $150 | Reasoning, Vision |
| xAI: Grok 4.3 | xAI | 1000K | $1.25 | $2.50 | Reasoning, Vision |
xAI: Grok 4.20 Multi-Agent currently leads on CoreAI with a 2000K-token context window. 50 models now reach the 1M-token mark, so there are several options for very long documents.
The context window is the maximum amount of text — measured in tokens — a model can consider at once, covering your prompt, attached files, the conversation so far, and its reply. Content beyond the window is invisible to the model.
Roughly 750,000 words — about 8 to 10 full-length novels, or a large software project's entire source code. That's enough to analyze complete books, long legal documents, or a year of meeting notes in one prompt.
No — it extends memory, not intelligence. Models can also lose precision in the middle of extremely long inputs, so for peak accuracy, keep critical instructions near the beginning or end of your prompt.
Chat with GPT-5, Claude, Gemini and hundreds more — all in one app.