Oct 27 7 days ago

1. 14:34

   feat: firela-researcher = Qwen2.5-72B-128K (TA context overflowed 14B's 32k) Running

   yuming
   4451142
2. 09:46

   fix: disable /responses bridging — route anthropic messages via chat/completions

   yuming
   13d86e0
3. 09:37

   fix: mode=chat on all models — litellm was bridging to OpenAI /responses (SiliconFlow 404)

   yuming
   0d0c617
4. 09:17

   fix: set litellm master key (new litellm refuses keyless startup)

   yuming
   3455f16
5. 08:26

   fix: worker model -> Qwen2.5-14B-Instruct (Coder-14B delisted from SiliconFlow)

   yuming
   c1791cc
6. 03:21

   feat: LiteLLM sidecar — Anthropic /v1/messages -> SiliconFlow for agents⁠

   yuming
   9b9c304