CRUST-bench: leaderboard
CRUST-Bench — C-to-safe-Rust transpilation benchmark of 100 real repositories, validated by Rust test suites.
Metric: Pass@1 Test Success (%). Source: crust-bench.github.io. Saturation forecast: Around July 2027. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 26 | #83 |
| 2 | GPT-5 (High) | 26 | #98 (GPT-5) |
| 3 | Gemini 3 Pro (Preview) | 23 | #66 |
| 4 | Claude Opus 4 (20250514) | 22 | #154 |
| 5 | O3 (2025-04-16) | 19 | #121 |
| 6 | O1 Preview (2024-09-12) | 15 | #197 |
| 7 | Claude 3.7 Sonnet (20250219) | 13 | #201 |
| 8 | Claude 3.5 Sonnet (20240620) | 11 | #299 |
| 9 | O1 Mini (2024-09-12) | 9 | #403 |
| 10 | GPT-4o | 7 | #343 |
| 11 | Gemini 1.5 Pro | 3 | #445 |
| 12 | QwQ 32B-Preview | 0 | #651 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=crust-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.