Claude Code v2.1.293 ·
Model list entries can cap the highest effort level offered
Each model in the model list can now set maxEffort, the highest reasoning effort level offered for it
You'll noticeTier: how much it should matter to you
4Useful: my rating, 1 to 5
3Signal: worth watching, 1 to 5
EffortArea: what it touches
ImprovementsKind: in v2.1.293,
What probably matters to youSection of the release
Unclear It is not clear whether the cap is applied anywhere other than Claude Desktop.
What
Each entry in the model list can now carry a maxEffort setting. Effort is how much reasoning the model is asked to do. The setting works like this:
- Effort levels above
maxEffort are hidden and never requested by Claude Desktop.
- An unrecognised value caps the model at low.
- When two spellings of the same model with 1M-token context are merged, the lower
maxEffort is kept.
Why
Administrators can limit how much reasoning effort each model is allowed to use.
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtIt is not clear whether the cap is applied anywhere other than Claude Desktop.
See this entry in the whole of v2.1.293 →
