Models and tiers
Dooze runs local models on your Mac and connects to cloud models from OpenAI, Anthropic, Google, and xAI.
Local models
On-device models run directly on your Mac using MLX. No internet required. These are available on all plans, including Free. See On-device models.
Cloud models
Cloud models are faster and more capable for complex tasks. Two types:
Fast models respond quickly. Good for everyday tasks.
Reasoning models think deeper. Better for complex problems, analysis, and longer outputs.
How to switch
Click the model name in the bottom bar of the panel. The dropdown shows available models, each with a colored dot for the provider and an icon for the type. You can switch between local and cloud per conversation.
Three plans
| Plan | Price | What you get |
|---|---|---|
| Free | $0/mo | On-device models, unlimited transcription, meeting recording, skills |
| Lite | $12/mo | OpenAI, Anthropic, xAI, Gemini models, natural voices, text polish, $10 AI credits/mo |
| Heavy | $25/mo | State-of-the-art models (Opus, GPT-5.4), $22 AI credits/mo, early access, priority support |
All plans include unlimited on-device features. No credit card required for Free.
To upgrade, go to Settings > Account.
AI credits
Lite and Heavy plans include monthly AI credits that cover cloud model usage, cloud TTS voices, and cloud reprocessing. Usage is tracked in Settings > Account.
Web search
Some conversations benefit from current information. Three modes: Auto (Dooze decides when to search), Required (always search), Off (never search). Toggle this in the model dropdown.
If you are unsure
Start with Free. Local models handle most everyday tasks. Upgrade when you want cloud model access or natural voices.