troubleshooting local AI
What the error messages mean, where files live on your machine, and how to restart a stuck engine.
yapyap tries to give you a specific reason rather than a vague "something's off". Here's what the messages you might see actually mean.
messages and their meaning
"… isn't in your Ollama library or model folders — start Ollama or pick another model" You picked one of your own Ollama models, Ollama isn't running, and yapyap can't find that model on disk anymore (it was moved or removed). Start Ollama, re-add the folder under model folders, or pick one of the built-in models.
"… this model needs Ollama running" This model is stored in a format yapyap's built-in engine can't run — only Ollama can. Start Ollama to use it, or pick a built-in model.
"request … exceeds the available context size …" The transcript plus the lens is longer than this model can take in at once. Pick a model that handles more (Qwen handles the most) or run the lens on a shorter recording.
"the engine didn't start …" The built-in engine couldn't start, and the message carries its own last words. With your own models this is usually a model too new for the built-in engine to understand — those keep working through Ollama.
a download seems stuck Model downloads show progress and can be cancelled from the picker. Cancelling keeps your previous model selected.
where things live on your machine
| what | where |
|---|---|
| models yapyap downloaded | <app data>/models/ |
| the built-in engine | <app data>/bin/ |
| your Ollama library | see using your own models |
yapyap never writes into your Ollama library — it only ever reads from it.
restarting the engine
- stop it — the insight settings page has a stop control; the engine also stops itself after about 15 minutes idle.
- start fresh — switching model restarts it automatically, and quitting yapyap always shuts it down (it never keeps running after the app closes).
- re-download a model — delete it from
<app data>/models/and pick it again.