Choosing and switching AI models
Last updated March 18, 2026
Typeahead ships with three models. You choose one during onboarding and can switch at any time from Settings.
The available models
- Gemma 3 4B (2.5 GB): Recommended for most users. Strong suggestion quality, good speed on all Apple Silicon Macs.
- Qwen 3 4B (2.5 GB): Same size as Gemma 3. Different training data. Some users prefer its completions for technical writing or structured text.
- Qwen 3 1.7B (1.7 GB): The lightest option. Faster inference, uses less memory. Better suited to Intel Macs or Macs with 8 GB of RAM.
How to switch models
Click the Typeahead icon in the menu bar and go to Settings, then Model. Choose a different model from the list. If you have not downloaded it yet, Typeahead will prompt you to download it first.
The new model loads automatically after the download. No restart required.
Can I have multiple models downloaded?
Yes. Each model is stored separately on disk. Switching between them is instant if they are already downloaded. Only one model runs at a time.
Where are the model files stored?
Models are stored in your Application Support folder. You do not need to manage these files manually. If you want to free up space, switch to a lighter model and remove the unused one from Settings > Model.
Still have questions?
Contact Support