Pi v0.80.10: Kimi Coding adaptive thinking fixes
Patch release fixing Kimi Coding model thinking compatibility — adaptive thinking now works correctly, K3 exposes its correct max thinking level, and empty-signature thinking blocks can be replayed. Also restores xAI models that were dropped in v0.80.9.
PUBLISHED2026-07-16
OBSERVED2026-08-21
AGE1mo
SOURCES1
- Kimi Coding models now use adaptive thinking correctly; K3 exposes only its supported
maxlevel - Empty-signature thinking blocks are replayable for K3 and kimi-for-coding
- Fixed pricing metadata for Moonshot AI and Moonshot AI China
- Restored xAI models in catalog generation (removed in 0.80.9)
COMMUNITY
No curated reactions recorded for this event. Facts and takes are kept in separate layers — community context is added by hand, never blended into the record above.