By John Gruber
CoverSutra Is Back from the Dead — Your Music Sidekick, Right in the Menu Bar
Google:
Gemini is also our most flexible model yet — able to efficiently run on everything from data centers to mobile devices. Its state-of-the-art capabilities will significantly enhance the way developers and enterprise customers build and scale with AI.
We’ve optimized Gemini 1.0, our first version, for three different sizes:
- Gemini Ultra — our largest and most capable model for highly complex tasks.
- Gemini Pro — our best model for scaling across a wide range of tasks.
- Gemini Nano — our most efficient model for on-device tasks.
Loosely speaking, Gemini Ultra is competing with GPT 4, and Gemini Pro with GPT 3.5. Nano, the on-device model, will first appear on Pixel 8 Pro phones. It’s unclear to me whether that’s because Gemini Nano is tuned to specifically take advantage of the Pixel 8 Pro’s Tensor G3 chip, or if it will expand to additional Android phones with other silicon.
Google has a 6-minute demo of Gemini in action, and it’s rather incredible. But it also comes with this disclaimer: “For the purposes of this demo, latency has been reduced and Gemini outputs have been shortened for brevity.” Why not show it in real time, even if it’s slow? It seems like the whole demo ought be considered fraudulent — a fake. What’s wrong with Google as a company that they repeatedly try to pass off concept videos as legitimate demos of actual products?
★ Thursday, 7 December 2023