← all entries
A-TIER tool

React — → http://github.com/ruucm/shadergradient

michaelpkocher · 2026-05-11T18:43:37.000Z · in AI Saves, Vol. 7
Open post ↗

100k ft — What is this?

React — → http://github.com/ruucm/shadergradient
toolA-tier

50k ft — Why care?

The most insane 3D, liquid, gradient & more UI effects you’ve seen are free on GitHub. This truly is your new design ... — @fifbuilds; uses React.
3DLLMUI/UXarchitecturedesignopen source

10k ft — Key takeaways

(no takeaways yet)

5k ft — Tools & techniques

React (no techniques tagged)

1k ft — Deep-dive / try it

The most insane 3D, liquid, gradient & more UI effects you’ve seen are free on GitHub. This truly is your new design asset library — and now that you can vibe code, you can use them too. 🔗 REPOS: ShaderGradient → http://github.com/ruucm/shadergradient liquid-logo → http://github.com/paper-design/liquid-logo liquid-glass-js → http://github.com/dashersw/liquid-glass-js react-three-fiber → http://github.com/pmndrs/react-three-fiber #aitools #productdesign #webdevelopment #github #3ddesign fernando_0422   6h hii can you send me the github repo :)
Transcript excerpt
Someone just built a large language model that's o and log o instead of o and squared, and it's within 5% of transformer quality. If you don't speak computer science, here's what that means. Current models get exponentially more expensive as the context window grows. This new architecture Log rhythmic cost. That's the difference between $400 a month and $40 a month at scale. It uses wave equation dynamics instead of self-attention. Sounds academic until you realize what this actually unlocks. You can process 100,000 word documents for the same cost as most platforms charge 10,000 words. The repo just went up a few days ago, and it's not quite in production yet But if you're building AI features with large context requirements, this is the breakthrough that you've been waiting for. Follow for technical shifts That actually changed the economics, not just the benchmarks.

Collection: AI Saves, Vol. 7