Set Up Providers
A provider tells NextClaw which model service to call. For the first setup, do not try to design the perfect stack. Choose the path most likely to work quickly.
Recommended order
- If you want the fastest path, start with Pick a Provider Path.
- If you already have an API key, configure that provider.
- If you want a local model, use Local Ollama + Qwen3.
Minimum setup
- provider name
- API base or platform entry point
- authentication method
- default model
After saving, return to the UI and send one real message.
When to add multiple models
Do not start by configuring every model you can find. Multiple models make sense when:
- one model is for fast drafts
- one model is for harder reasoning
- one model is local or offline
- different sessions need different bindings
Common issue
The model is configured but no reply arrives
Run:
bash
nextclaw doctorThen check:
- whether the API key is valid
- whether the model name exists for that provider
- whether the default model was saved