NVIDIA's AVO Agent Scores 100% on ARC-AGI-3 Benchmark

eigenhector · x · 2026-08-21

NVIDIA's research coding agent, AVO, achieved a perfect 100% score on the ARC-AGI-3 interactive reasoning benchmark, completing all 183 levels. Notably, AVO was not designed for this benchmark and succeeded with minimal input and tooling by analyzing past grids represented as text.

Related event: NVIDIA's AVO coding agent scores 100% on ARC-AGI-3(7 posts)→

Original post →

More from coding & agent

coding & agent channel →