CTFBench: leaderboard
CTFBench: Measures model robustness, truthfulness, calibration, bias, harmfulness, jailbreak resistance, or alignment-relevant behavior.
Metric: Vulnerability Detection Rate (imported). Source: github.com. Source: benchmarklist.com. Saturation forecast: Around July 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 100 | #31 |
| 2 | Gemini 3.1 Pro (Preview) | 98.4 | #58 |
| 3 | Claude Opus 4.6 | 88.9 | #63 |
| 4 | Kimi K2.6 | 82.5 | #104 |
| 5 | Claude Opus 4.7 | 73 | #49 |
| 6 | GPT-5 | 71.4 | #98 |
| 7 | Claude Opus 4.5 | 71.4 | #83 |
| 8 | DeepSeek V4 Pro | 64.8 | #102 |
| 9 | Gemini 2.5 Pro | 57.1 | #152 |
| 10 | GPT-5.4 | 57.1 | #79 |
| 11 | MiMo-V2.5-Pro | 55.6 | #158 |
| 12 | GPT-5.2 | 52.4 | #110 |
| 13 | MiniMax-M2.7 | 50.8 | #258 |
| 14 | DeepSeek R1 | 42.9 | #251 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ctfbench · How It Works · Data refreshed daily, snapshot 2026-10-20.