Nvidia says its general-purpose coding agent system AVO scored 100% across all 25 environments in the ARC-AGI-3 public set, completing all 183 levels (Terry Chen/NVIDIA Technical Blog)
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Nvidia's AVO scored 100% across all 25 environments in the ARC-AGI-3 public set, completing all 183 levels
- AVO elevates Claude Opus 5 from a 30% model baseline to 100% as part of its complete agent system
Why it matters: AVO lifted Claude Opus 5 from 30% to a perfect score on ARC-AGI-3, showing that agent scaffolding around a frontier model can multiply performance on novel reasoning tasks — the gains came from the system wrapping the model, not from the underlying model itself.
Ask SkimNews
