Automated Vulnerability Repair on AIxCC CP Dataset 1.0 (final round)
91.3Overall Success RatePatchIsland
Evaluation Results
| Method | Links | ||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PatchIslandLLMs=GPT o4-mini, Claude Sonnet 4, Gemini 2.5 Pro2026.01 | 91.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Buttercupconfiguration=Default AIxCC Submission2026.01 | 62 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RoboDuckconfiguration=Default AIxCC Submission2026.01 | 17.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |