Local Model Backend Not Running | Generated by AI
The error is clear: the local model backend isn’t running.
The system tried to call openai/gpt-oss-120b at http://localhost:1234/api/v1/chat and got Connection refused — there’s no model server listening on port 1234.
Per the README:
A local OpenAI-compatible chat backend reachable at
http://localhost:1234/api/v1/chat
You need to start a local inference server (e.g. LM Studio, Ollama, llama.cpp server, vLLM) that:
- Listens on
http://localhost:1234 - Has the path
/api/v1/chat(OpenAI-compatible) - Serves the configured models (
qwen3.5-9b-mlx,openai/gpt-oss-120b, etc.)
Quick check — do you have LM Studio or Ollama installed? If so, start it and load a model, then the session creation will work.