DEVOURED
Schema
Schema is a new harness that enables AI agents to reach 99% accuracy on the ARC-AGI-3 benchmark by turning game observations into executable programs.
What: Developed to improve agentic reasoning, Schema forces models to translate visual game inputs into code, test those programs against the environment, and iteratively refine their plans.
Why it matters: This move toward 'executable reasoning' demonstrates that agents perform significantly better when they treat abstract problems as software development tasks rather than relying purely on pattern matching.
Decoder
- ARC-AGI-3: The Abstraction and Corpus Reasoning benchmark, a series of tasks designed to test general intelligence by requiring the model to solve novel logic puzzles with minimal examples.
Original article
Schema is a harness that makes frontier models play like physicists. It can help frontier models achieve 99% on ARC-AGI-3 Public. ARC‑AGI‑3 gives agents a game environment without an explanation of what it is seeing. Schema makes agents write each game's mechanism as an executable program, test it against reality, and plan inside it. It controls how observations are turned into a working model of a game, how predictions are tested against the interaction history, and how plans are executed and revised.
Explore all 54 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi