May 6, 2025
Two things here:
1. It's increasingly clear that the future of front-end software is that basically everything you see and interact with will be generated on the fly by LLMs.
2. Video understanding is a big deal. I did a livestream last week and I wanted a cleaned up version of everything I said, mapped to the slides that were on screen at the time. 5 minutes with the Gemini API and I had exactly what I wanted. (code in reply)
🙌 So stoked to share the latest Gemini model with you all today!
🥇#1 on the webdev arena
📽️ SOTA on video understanding
🖼️ beautiful UI visualizations
🤔 all of Gemini's reasoning and long context input / output
...and can even make you digital pets, like this cute example

Gist[1]