PetriBench - Medium: leaderboard
Exact-match accuracy on PetriBench's Medium tier, generated by increasing the structural complexity of the Petri nets behind the Easy questions. Reasoning effort is part of the row identity. Higher is better. Source: arxiv.org/abs/2609.19883.
Metric: Exact-Match Accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 36 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (xHigh) | 90.9 | #22 (GPT-5.6 Sol) |
| 2 | Claude Opus 5 (xHigh) | 83.8 | #16 (Claude Opus 5) |
| 3 | Claude Opus 5 (High) | 81.9 | #16 (Claude Opus 5) |
| 4 | GPT-5.6 Sol (High) | 81.4 | #22 (GPT-5.6 Sol) |
| 5 | Grok 4.6 (xHigh) | 80.9 | #29 (Grok 4.6) |
| 6 | Grok 4.6 (High) | 76.5 | #29 (Grok 4.6) |
| 7 | Grok 4.6 (Medium) | 75.7 | #29 (Grok 4.6) |
| 8 | Gemini 3.8 Flash (High) | 74.5 | #30 (Gemini 3.8 Flash) |
| 9 | Claude Opus 5 (Medium) | 70 | #16 (Claude Opus 5) |
| 10 | Gemini 3.1 Pro (Preview) (High) | 69.9 | #64 (Gemini 3.1 Pro (Preview)) |
| 11 | Grok 4.5 (High) | 67.2 | #54 (Grok 4.5) |
| 12 | Grok 4.5 (Medium) | 65 | #54 (Grok 4.5) |
| 13 | GPT-5.6 Terra (xHigh) | 64.4 | #40 (GPT-5.6 Terra) |
| 14 | Gemini 3.8 Flash (Medium) | 58.7 | #30 (Gemini 3.8 Flash) |
| 15 | GPT-5.6 Sol (Medium) | 52.5 | #22 (GPT-5.6 Sol) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=petribench-medium · How It Works · Data refreshed daily, snapshot 2026-11-05.