PetriBench - Hard: leaderboard
Metric: Exact-Match Accuracy (%). Source: arxiv.org. 36 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Sol (xHigh) | 77.8 |
| 2 | Claude Opus 5 (High) | 71.8 |
| 3 | Claude Opus 5 (xHigh) | 71.7 |
| 4 | GPT-5.6 Sol (High) | 59.8 |
| 5 | Claude Opus 5 (Medium) | 58.1 |
| 6 | Grok 4.6 (xHigh) | 56.5 |
| 7 | Grok 4.6 (High) | 55 |
| 8 | Grok 4.6 (Medium) | 52.6 |
| 9 | GPT-5.6 Terra (xHigh) | 50.7 |
| 10 | Grok 4.5 (High) | 50.1 |
| 11 | Grok 4.5 (Medium) | 48.9 |
| 12 | Gemini 3.1 Pro (Preview) (High) | 46.9 |
| 13 | Gemini 3.8 Flash (High) | 41.3 |
| 14 | Gemini 3.8 Flash (Medium) | 38.5 |
| 15 | Gemini 3.1 Pro (Preview) (Medium) | 34.8 |
Interactive version: theaggregate.ai/benchmark?slug=petribench-hard · How It Works · Data refreshed daily, snapshot 2026-09-20.