Vellum - ARC-AGI-2 — leaderboard
Vellum's independently-run ARC-AGI-2 evaluation. Abstract pattern recognition and reasoning tasks.
Metric: Score (%). Source: www.vellum.ai. Status: saturation imminent. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 85 |
| 2 | Gemini 3.1 Pro (Preview) | 77.1 |
| 3 | Gemini 3.5 Flash | 72.1 |
| 4 | Claude Opus 4.6 | 68.8 |
| 5 | Claude Sonnet 4.6 | 58.3 |
| 6 | GPT-5.2 | 52.9 |
| 7 | Claude Opus 4.5 | 37.6 |
| 8 | Gemini 3 Pro | 31 |
| 9 | GPT-5 | 18 |
| 10 | GPT-5.1 | 18 |
| 11 | Grok 4 | 16 |
| 12 | Kimi K2.5 | 12 |
| 13 | Claude Opus 4.1 | 9 |
| 14 | O3 | 7 |
Interactive version: theaggregate.ai/benchmark?slug=vellum-arc-agi-2 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.