CrypFormBench: leaderboard
CrypFormBench (2026) tests whether LLMs can carry out formal security analysis of cryptographic schemes in the specification languages of seven verification tools (Scyther, Tamarin, AVISPA, ProVerif, Maude-NPA, CryptoVerif and EasyCrypt) over 700 instances. The overall score (0 to 100) weights five task scores: 0.15 interpretation, 0.25 generation, 0.20 completion, 0.25 transformation and 0.15 correction. Higher is better.
Metric: Overall score (0-100): 0.15 interpretation + 0.25 generation + 0.20 completion + 0.25 transformation + 0.15 correction task scores on CrypFormBench (700 instances, 677 cryptographic schemes, 7 formal verifier languages); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 35 | #741 |
| 2 | DeepSeek R1 | 34.6 | #473 |
| 3 | GPT-4o Mini | 23 | #1623 |
| 4 | Gemini 2.5 Pro | 22.8 | #266 |
| 5 | Grok 3 | 18 | #602 |
| 6 | GLM-4 | 16.1 | #1666 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=crypformbench · How It Works · Data refreshed daily, snapshot 2026-11-13.