Manage downloaded models
Outcome
Section titled “Outcome”You can inspect and manage the local Whisper and, in Local Mode, Ollama models that Synthezia uses.
- Open Models in the sidebar or press Cmd+M.
- Under Transcription, select an installed Whisper model as the default or select an available one to download it.
- Under Summarization & Chat, select an installed Ollama model or expand Download more models to pull an additional catalog model.
- Use Refresh after a completed or failed local Ollama model operation.
- Use the delete control only for a model you no longer need, then confirm the deletion. It can be downloaded again later when available.
Expected result
Section titled “Expected result”The model inventory reflects installed models and their active local selections. Synthezia prompts for a required model instead of starting a dependent local workflow without one.
Network and storage
Section titled “Network and storage”Whisper model downloads use the whisper.cpp model repository on Hugging Face unless a packaged copy is available. Ollama model downloads are requested through the local runtime from its configured registry. Both store models in the application data area.
Common problems
Section titled “Common problems”- A pull or download fails: check network access, retry, and refresh the inventory before starting the dependent workflow again.
- Ollama controls are absent: choose Local (Ollama) in Settings; External API does not use local Ollama models for summaries and chat.
- You deleted a model accidentally: download it again if it remains available.