Model not loading

Soriku says the model isn't available, or the request hangs without a response. Five quick checks.

1. Is Ollama running?

curl -s http://localhost:11434/api/tags | jq '.models[].name'

Empty response or connection refused means Ollama isn't up. Start it with ollama serve or relaunch the app.

2. Is the model pulled?

ollama list

Model not in the list? Pull it: ollama pull qwen2.5-coder:7b. Soriku doesn't auto-pull because pulls can be multi-GB.

3. Right model name?

Ollama is strict about model tags. qwen2.5-coder:7b and qwen2.5-coder are different. Soriku follows Ollama's naming. If you pulled qwen2.5-coder:latest, ask for that exact tag.

4. Enough RAM?

A 7B model needs ~5 GB free. If you already have a model loaded and not enough headroom, the load silently fails. See Memory management.

5. Does Soriku see Ollama?

curl -s http://localhost:8765/api/v1/models | jq '.data[] | select(.provider == "ollama") | .id'

If Ollama models are missing from Soriku's list while ollama list shows them, restart Soriku. The auto-discovery runs at startup.