CAM-Bench: leaderboard
Formal theorem-proving benchmarks enable mechanically verifiable evaluation of mathematical reasoning in large language models. However, existing benchmarks mainly focus on Olympiad-style problems and algebraic domains, leaving computational and applied mathematics underrepresented. We introduce CAM-Bench, a Lean 4 theorem-proving benchmark of 1,000 Lean proof targets in computational and applied
Metric: Pass@32 (self-reported). Source: benchmarklist.com. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4 Pro | 19.67 | #102 |
| 2 | Claude Sonnet 4.6 | 19.28 | #90 |
| 3 | Gemini 3.1 Pro (Preview) | 18.38 | #58 |
| 4 | GPT-5.4 | 11.7 | #79 |
| 5 | Kimi K2.6 | 2.25 | #104 |
Interactive version: theaggregate.ai/benchmark?slug=cam-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.