Platform
Reasoning
A model router tuned per tier, never exposed as a raw API choice.
Overview
You work with named Paw models (Flash, Swift, Core, Vision, Voice and Memory) matched to your tier — nothing to set up or manage.
How it works
- 1
You ask
A request comes in as text, voice, or an uploaded image.
- 2
Route
The right Paw model handles it based on the task and your tier.
- 3
Respond or act
The model either answers directly or calls a real tool to take action.
Use cases
- • Getting the right capability for the job without picking a model yourself
- • Vision-based tasks like recreating a design from a screenshot
Limitations
- • Reasoning quality scales with your tier's model access — Go includes Paw Flash for planning and analysis; execution requires Pro