You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This commit was created on GitHub.com and signed with GitHub’s verified signature.
Changed
New default model names for the LiteLLM router: the router groups qwen and bge-m3 are
being retired. CLEANUP_LLM_MODEL (server) and AI_REM_LLM_MODEL (client) now default to private-general (local-ai on mystudio, qwen3-14b as fallback), EMBED_HTTP_MODEL defaults to embed (the same bge-m3, 1024 dimensions — no re-embedding needed). An explicit value in the
deployment .env still wins; the router key must allow the new names before deploying (#211).