Agent task performance on AgentBench DB-Bench 20-task (held-out)
54.7G ScoreOffline AW Controller
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Offline AW ControllerHarness configuration=AW, Scoring Protocol=Adapted Harness MDP2026.07 | 54.7 | 16.7 | |
| Base HarnessHarness configuration=Base, Scoring Protocol=Adapted Harness MDP2026.07 | 41.5 | — |