AI & Machine Learning

GPT-5, Claude, Gemini All Score Below 1% - ARC AGI 3 Just Broke Every Frontier Model

Ali NematiAli Nemati6 hours ago30 sec read8 views

ARC-AGI 3, launched on March 25, 2026, introduces interactive game environments where AI agents must discover rules and solve problems without internet access, marking a significant shift from static benchmarks. Current frontier LLMs perform poorly in this format, achieving less than 1%, while simple RL and graph search methods reach up to 12.58%. This competition highlights the need for novel algorithmic ideas over model scaling, with $2 million in prizes incentivizing open-source solutions.

Read the full article at DEV Community


Want to create content about this topic? Use Nemati AI tools to generate articles, social posts, and more.

8
Comments
Ali Nemati
Ali NematiWritten by Ali
View all posts

Related Articles