AI assistants now run on Claude Sonnet 5.5 and Opus 5.5
Adrian Duyzer
The in-app AI assistants now run on Claude Sonnet 5.5 by default, with Claude Opus 5.5 available from the model dropdown, replacing Sonnet 5 and Opus 4.8.
Impact: you do not have to do anything. Every assistant that used Sonnet 5 now uses Sonnet 5.5, and anyone who picked Opus 4.8 from the dropdown will find Opus 5.5 in its place.
- The model dropdown appears in the mapping assistant, the source and semantics assistants, and trace diagnostics. It lists Sonnet 5.5 and Opus 5.5.
- The help assistant uses Sonnet 5.5 when you are signed in.
- The mapping assistant on the public demo uses Sonnet 5.5, and you cannot change it there.
- Opus 5.5 reasons before it answers, which Opus 4.8 did not, so expect Opus responses to take longer than they used to.
- Every assistant now has a 64,000-token output budget. Previously only the mapping assistant did, and the others stopped at 16,384, so a long answer from another assistant is less likely to be cut off.
How to choose a model for the mapping assistant is in the mapping docs.
This update was written with AI assistance. It has been reviewed and edited by a human.