From 0% to 36% on Day 1 of ARC-AGI-3
Summary
Symbolica's Agentica framework reports 36.08% on ARC-AGI-3 Day 1, outperforming CoT baselines with a lower cost. The post includes a gallery of game results, a full score breakdown, a chat with Agentica, and a link to the GitHub repository, providing actionable data for researchers and SMBs exploring agent-based AI and automation.