Watch three of the most advanced AIs on Earth attempt Leonardo’s greatest work Mona Lisa
Fable 5 vs Opus 5 vs GPT-5.6 Sol
Claude Opus 5 scores around 30% on ARC-AGI-3 by itself.
NVIDIA wrapped the exact same model in their AVO agent system. Result: 100%. All 183 levels. All 25 environments. Zero instructions. Zero stated goals. The model didn’t improve. The architecture around it did. This is
Hermes plus Kimmy K3 is the best AI agent combo right now.
An open model just beat the biggest names at coding. Kimmy K3 went straight to #1 on the front-end arena. It beat Fable 5, GPT 5.6, and Opus 4.8. But most people use it the weakest way possible: typing into a chat box.
Claude Opus 5 is INSANE
Look what I’ve just built.
You have reached the end of the archive
All of Claude Opus 5