Claude Opus is reported to have a 1 million-token context window in its latest 4.6 release, up from the earlier 200,000-token window in 4.5. That means it can handle very large prompts, such as long documents or substantial codebases, in a single conversation.

Quick scoop

For the Claude app specifically, the practical limit you hit can depend on the exact model version and product surface, but Opus 4.6 is the big headline: 1 million tokens of context in beta on the developer platform at launch. Anthropic’s own release for Opus 4.6 also positions it as the newest Opus- class model as of February 2026.

What that means

  • 200,000 tokens is already a large context window, roughly several hundred pages of text.
  • 1 million tokens is much larger and is meant for tasks like massive code review, long documents, or extended agent workflows.
  • In real use, the app may still apply product-specific limits, so “model context” and “what you can paste in the app” are not always identical.

Simple answer

If you mean Claude Opus 4.6 , the context is 1 million tokens. If you mean a specific Claude app plan or interface, the exact usable limit can vary by product setup.

TL;DR: Opus has 1M tokens in its latest reported release, while earlier Opus 4.5 was 200K.