NVIDIA Says Its AVO Coding Agent Cleared Every Public Level of the ARC-AGI-3 Reasoning Benchmark
Sign in✓ Signed in
NVIDIA says AVO, a general-purpose coding agent built by the company, has cleared all 183 levels across the 25 public environments of ARC-AGI-3, an interactive reasoning benchmark designed specifically to resist automated agents by withholding instructions, rules, or stated goals. According to NVIDIA, the agent had to figure out what to do in each environment purely through exploration and trial and error, with no natural-language guidance of any kind.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...