Channels
Engine
Google's low-latency model. Plans its approach in a first pass, then writes the answer in a second — both passes are timed in the transcript.
A reasoning-first model. Works through problems step by step, so it trades speed for stronger logic and multi-step maths.
Meta's compact open model. Answers in a single pass with no planning stage, which makes it the quickest of the three.
Navigation
Listening...