benchmarks evals
A Visual Harness Reached Perfect Action Efficiency
Lossless observation memory let one multimodal model finish all 25 public ARC-AGI-3 games.
Summary
Lossless observation memory let one multimodal model finish all 25 public ARC-AGI-3 games.
VISTA preserves past visual observations in original form and lets a multimodal model retrieve and reorganize them during long-horizon reasoning. On ARC-AGI-3, the authors report that it raised Claude Opus 5.0's Relative Human Action Efficiency from 40.68 to 100, completing all 25 public games with 57.4 percent fewer actions than first-time human participants. It also improved results on three other visual-game and puzzle benchmarks. The score depends on the reported harness, model and public game set and is not a general measure of visual intelligence.
Why it matters
Lossless observation memory let one multimodal model finish all 25 public ARC-AGI-3 games.
Limits and context
- The score depends on the reported harness, model and public game set and is not a general measure of visual intelligence.
Key claims
Lossless observation memory let one multimodal model finish all 25 public ARC-AGI-3 games.
Qualification: The score depends on the reported harness, model and public game set and is not a general measure of visual intelligence.
Evidence: source-2026-10-02-006
Sources
- arXiv preprint 2610.02200arXiv · primary research
Corrections
No corrections have been recorded for this story.