AutomataBench: leaderboard
AutomataBench — constructing and reasoning about finite automata and formal languages; weighted pass@1 across task tiers.
Metric: Weighted pass@1 (%). Source: www.automatabench.com. Saturation forecast: Around April 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 45.28 | #31 |
| 2 | Claude Opus 4.8 | 30.22 | #37 |
| 3 | DeepSeek V4 Pro | 28.34 | #102 |
| 4 | Gemini 3.1 Pro (Preview) | 14.7 | #58 |
| 5 | Grok 4.3 | 6.51 | #188 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=automatabench · How It Works · Data refreshed daily, snapshot 2026-10-20.