CyberGym-E2E S4 (920 tasks)
Agents · 2026-06-03
CyberGym-E2E cumulative S4 success rate on the current 920-task set. A task must pass S1 through S3 and the patch must also fix the specific held-out ground-truth vulnerability.
Top models (higher is better)
| Model | Score |
|---|---|
| Opus 4.6 | 26.2 |
| GPT-5.4 | 22.2 |
| Gemini 3.1 Pro Preview | 20.5 |