ATOBench - SQLi (Native): leaderboard
Metric: Grounded verification rate (%; the direct-exploit-proof contract; over ATOBench's 15 matched penetration-testing episodes per contract; requires registered task evidence, positive report closure and trace support; native target (no adversarial response transformation)). Source: arxiv.org. Saturation forecast: Around August 2027. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (High) | 60 |
| 2 | DeepSeek V4 Pro (High) | 60 |
| 3 | GLM-5.2 (High) | 46.7 |
| 4 | Kimi K2.6 (High) | 26.7 |
Interactive version: theaggregate.ai/benchmark?slug=atobench-sqli-native · How It Works · Data refreshed daily, snapshot 2026-09-26.