CMT-Benchmark: leaderboard
Metric: Pass@1 (%). Source: arxiv.org. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 30 |
| 2 | O3 | 26 |
| 3 | GPT-5 Mini | 24 |
| 4 | O4 Mini | 18 |
| 5 | Gemini 2.5 Pro | 14 |
| 6 | GPT-5 Nano | 14 |
| 7 | Llama 4 Maverick | 12 |
| 8 | Gemini 2.0 Flash | 10 |
| 9 | Claude Opus 4.1 | 8 |
| 10 | Claude 3.7 Sonnet | 6 |
| 11 | Claude Sonnet 4 | 6 |
| 12 | GPT-4.1 | 4 |
| 13 | Gemini 2.5 Flash | 4 |
| 14 | DeepSeek V3 | 4 |
| 15 | GPT-4o | 2 |
Interactive version: theaggregate.ai/benchmark?slug=cmt-benchmark · How It Works · Data refreshed daily, snapshot 2026-09-05.