Play Now
All patch notes

Patch Notes B0.041 - Model Changes

Model Changes

  • Added Muse Spark 1.3 back to the core lineup.

  • Added Mercury 2.5 to the core line up. It's very fast! It's currently on an 80% discount from the provider and will remain in core until the discount is over.

  • Added Gemma 4 31B Nitro, a very fast version of Gemma 4 31B, to the model picker. This is the fastest model we serve, but it's a bit pricey.

  • Removed GLM 5.3 Flash from core. Unfortunately it's just too slow and causing too many problems to be a core model.

Improvements and Fixes

  • Orbit can now delete an assistant instruction for you, after you approve it.

  • Added a copy button for the request ID on each row of your usage history, so you can include it when reporting a problem

  • Fixed Orbit background tasks on Muse Spark finishing their edits but reporting nothing back.

  • Improved the reliability of AI file edits so fewer find-and-replace changes fail and need a retry.