🚀 Grok-4 just snagged the #1 spot on ARC-AGI and ARC-AGI-2! 🎉
Here’s what the ARC-AGI team is saying: "Yesterday we got a call from xAI asking to test Grok-4. We’ve heard whispers it’s impressive, but we didn’t expect it to top the charts!"
Let’s break it down: Grok-4 is now the strongest publicly available model on ARC-AGI 1/2, even outshining some specialized solutions on Kaggle! 🏆
ARC-AGI-2 was designed for cutting-edge models, requiring them to master mini-skills on training sets and showcase them during testing. The previous high score hovered around 8% (looking at you Opus 4), but anything under 10% is just noise. 🤷♂️
Grok-4 crushed that noise with a whopping 15.9%, proving real flexible intelligence is here to stay! 💪✨ And the best part? It’s still competitively priced like Claude Opus! 💰
#AI #TechNews #Grok4
