Web Task Automation on WorkArena L1
68Average RewardJEF-HINTER
Evaluation Results
| Method | Links | |
|---|---|---|
| JEF-HINTERBase Model=GPT-5-mini2025.10 | 68 | |
| Human hintsBase Model=GPT-5-mini2025.10 | 66 | |
| DocumentationBase Model=GPT-5-mini2025.10 | 64 | |
| ReActBase Model=GPT-5-mini2025.10 | 61 | |
| JEF-HINTERBase Model=GPT-5-nano2025.10 | 48 | |
| DocumentationBase Model=GPT-5-nano2025.10 | 44 | |
| Human hintsBase Model=GPT-5-nano2025.10 | 43 | |
| ReActBase Model=GPT-5-nano2025.10 | 41 |