using your own models
Already have language models on disk? yapyap reuses them — your Ollama library and any extra folders you point it at — with or without Ollama running.
If you already have language models on your machine, yapyap finds and reuses them instead of downloading again.
your Ollama library — found automatically
yapyap reads your Ollama models straight from disk, so they're available even while Ollama isn't running — no setup needed. The usual locations are checked on every platform:
| platform | where Ollama keeps models |
|---|---|
| macOS | ~/.ollama/models |
| Windows | C:\Users\<you>\.ollama\models |
| Linux | ~/.ollama/models |
Models found there show up in the insight model picker marked Installed locally, alongside the built-in options. Pick one and it just works — Ollama runs it when it's up, and yapyap's built-in engine runs it straight from the same files when it isn't.
unknown license
The built-in options are chosen for licenses that are safe to use commercially. A model from your own library is yours to check — the picker marks it "Unknown license" and asks you to confirm before you pick it.
extra model folders
settings → insight → model folders (also under preferences) lets you point yapyap at other places you keep models — a folder of files you downloaded yourself, or a second Ollama library on an external drive or a NAS. Anything found there appears in the picker and works offline, just like your main library.
what isn't supported
- Loose model files that aren't in the picker. yapyap uses a file you've downloaded yourself only as the weights for a model it already offers, so a folder of other files never silently swaps your model.
- Formats the built-in engine can't read. A model stored in an unsupported format is only available while Ollama can run it — and if you pick it while Ollama is off, yapyap tells you so plainly.