CyberGym-E2E S2 (920 tasks)
Agents · 2026-06-03
CyberGym-E2E cumulative S2 success rate on the current 920-task set. A task passes when the agent first generates a crashing proof of concept and then patches the crash.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.4 | 66.2 |
| Opus 4.6 | 65.0 |
| Gemini 3.1 Pro Preview | 44.3 |