CTFBench — leaderboard
CTFBench: Measures model robustness, truthfulness, calibration, bias, harmfulness, jailbreak resistance, or alignment-relevant behavior.
Metric: Vulnerability Detection Rate (self-reported). Source: benchmarklist.com. Status: saturation imminent. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 100 |
| 2 | Gemini 3.1 Pro (Preview) | 98.4 |
| 3 | Claude Opus 4.6 | 88.9 |
| 4 | Kimi K2.6 | 82.5 |
| 5 | Claude Opus 4.7 | 73 |
| 6 | GPT-5 | 71.4 |
| 7 | Claude Opus 4.5 | 71.4 |
| 8 | DeepSeek V4 Pro | 64.8 |
| 9 | Gemini 2.5 Pro | 57.1 |
| 10 | GPT-5.4 | 57.1 |
| 11 | MiMo-V2.5-Pro | 55.6 |
| 12 | GPT-5.2 | 52.4 |
| 13 | Grok 3 | 52.4 |
| 14 | MiniMax-M2.7 | 50.8 |
| 15 | DeepSeek R1 | 42.9 |
Interactive version: theaggregate.ai/benchmark?slug=ctfbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.