The Context Window, Explained Simply
What Claude's context window is, how big it is in 2026, and why it matters for working with long documents and chats.
The context window is how much text Claude can hold in its head at once — your prompt, any files you upload, and the whole conversation so far. Everything you want Claude to reason over has to fit inside it. Think of it as Claude's short-term memory for the current chat.
How Big Is It?
| Model | Context window | Roughly |
|---|---|---|
| Opus 4.8 / Sonnet | 1,000,000 tokens | ~750,000 words — a small shelf of books |
| Haiku 4.5 | 200,000 tokens | ~150,000 words — a 500-page book |
A token is about ¾ of a word. In 2026 the top Claude models reached a 1-million-token window, and Anthropic removed the old surcharge for using very long context. In practice, that means you can drop in an entire contract, a whole codebase, or months of notes and ask questions across all of it — without chopping the material into pieces and losing the links between them.
Why It Matters to You
- Bigger window = you can give Claude more source material at once, so answers are grounded in your actual documents.
- Long chats stay coherent — Claude remembers what you said earlier in the same conversation.
- It's shared: a huge uploaded file leaves less room for a long back-and-forth. If a chat gets very long, start a fresh one.
Want the context to stick around across many chats instead of one? That's what Projects are for.