You'll noticeTier: how much it should matter to you
3Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
Context WindowArea: what it touches
ImprovementsKind: in v2.1.301,
ImprovementsSection of the release
Unclear Which models are affected is not stated.
What
The context window is how much conversation, code and other text a model can hold at once. Claude Code has reworked how it decides the size of that window:
When your opt-in to the 1-million-token window is honored, the window is 1 million tokens.
A model that supports 1 million tokens only as a beta, and not natively, is capped at the default size unless there is an opt-in or another override.
When Claude Code tidies up a model name, it adds the [1m] marker only if the model supports 1 million tokens natively.
Why
For some models this can change how much context is available and how soon Claude Code compacts the conversation. Compacting means summarizing older parts of the conversation to free up space.
How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtWhich models are affected is not stated.
