Common issues

Ollama won't connect

• Confirm Ollama is actually running before opening Trace or switching providers.
• Check that the model Trace expects (qwen3:8b by default) has been pulled locally.
• Make sure your machine meets the minimum requirements; Ollama needs enough free RAM to load a model.

"Invalid API key" for a cloud provider

• Double-check the key was copied in full, with no leading or trailing spaces.
• Confirm the key hasn't been revoked or rotated on the provider's own dashboard.
• Make sure you're pasting the key into the matching provider field. OpenAI, Anthropic, and Gemini keys aren't interchangeable.

The embedding health indicator shows a problem

This usually means the embedding model isn't reachable, most often because the active AI provider is unreachable for the same reason as above. Fixing the provider connection typically fixes this too. A manual re-check is available in Settings → AI.

Session restore isn't doing what I expect

Check Settings → Data for the session-restore preference: always restore, always start fresh, or ask every time. This setting is what decides the behavior, not a bug if it seems inconsistent.

OpenRouter shows as disconnected

OpenRouter uses an OAuth connection rather than a key. If it's showing disconnected, reconnect from Settings → AI rather than looking for a key field, there isn't one.

Still stuck?

Check the FAQ for setup and account questions that aren't strictly errors.