Autonomous Web Navigation on WebArena (refactored)
86.3Success RateEous (V)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Eous (V)Model=GPT-5, Accessibility Tree=false, Screenshot=true2026.01 | 86.3 | 4.7 | |
| Eous (S)Model=GPT-5, Accessibility Tree=true, Screenshot=false2026.01 | 75.3 | 4.2 | |
| AgentOccam (V)Model=GPT-5, Accessibility Tree=true, Screenshot=true2026.01 | 70.3 | 10.4 | |
| AgentOccam (S)Model=GPT-5, Accessibility Tree=true, Screenshot=false2026.01 | 69.2 | 9.4 | |
| MidSceneModel=Seed-1.8, Accessibility Tree=false, Screenshot=true2026.01 | 61.5 | 13 | |
| WebarenaBaseModel=GPT-5, Accessibility Tree=true, Screenshot=false2026.01 | 26.4 | 10 |