Skip to content

Manage models

Item Current behavior
Default Background model preparation is off; default Whisper and local Ollama models are selected separately.
Valid values The current in-app Whisper and Ollama catalog, plus the local embedding model required for search.
Existing sessions Existing records remain. A deleted model prevents dependent future local actions until it is restored.
New actions Use the selected installed defaults or show a required-model dialog.
Storage Model files are stored in the application-data models/ and ollama/models/ directories; selections are in webview local storage.
Privacy Model pulls require network access through their configured download paths. Local inference remains on device after models are present.
Restart None after a successful pull; refresh the inventory after an operation.
  1. Open Models with Cmd+M.
  2. Select or download a Whisper model under Transcription.
  3. In Local Mode, select or pull an Ollama model under Summarization & Chat.
  4. Use Refresh after an operation; delete only a model you are prepared to download again.

Enabling Prepare local AI models in the background in Settings permits optional downloads of the current recommendations. It does not make those downloads work without network access.