Vulnerability Discovery on TerminalBench 2 snapshot 2026-04-17
84.3Score (%)AgentFlow
Evaluation Results
| Method | Links | |
|---|---|---|
| AgentFlowBackbone=Claude Opus 4.6, Date=2026-04-172026.04 | 84.3 | |
| ForgeCodeBackbone=Claude Opus 4.6, Date=2026-03-122026.04 | 81.4 | |
| CapyBackbone=Claude Opus 4.6, Date=2026-03-122026.04 | 77.7 | |
| Terminus-KIRABackbone=Claude Opus 4.6, Date=2026-02-222026.04 | 77.3 | |
| Meta-HarnessBackbone=Claude Opus 4.6, Date=2026-03-302026.04 | 76.4 | |
| TongAgentsBackbone=Claude Opus 4.6, Date=2026-02-222026.04 | 74.6 | |
| DroidBackbone=Claude Opus 4.6, Date=2026-02-052026.04 | 72.4 | |
| MuxBackbone=Claude Opus 4.6, Date=2026-02-132026.04 | 69 | |
| CruxBackbone=Claude Opus 4.6, Date=2026-02-232026.04 | 66.9 | |
| Terminus 2Backbone=Claude Opus 4.6, Date=2026-02-062026.04 | 65.6 | |
| Claude CodeBackbone=Claude Opus 4.6, Date=2026-02-072026.04 | 60.9 |