Web Automation on WebArena
53.7Success RateSKILL.nb
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SKILL.nbEvaluation Protocol=single-round, Task Initialization=starts without task-specific persistent state2026.06 | 53.7 | 52 | 56.2 | 70.3 | 54.4 | 50.8 | 17.5 | — | — | — | — | — | — | — | |
| ReasoningBankEvaluation Protocol=single-round, Task Initialization=starts without task-specific persistent state2026.06 | 49.8 | 52.1 | 55.8 | 68.9 | 45.2 | 42.1 | 11.3 | — | — | — | — | — | — | — | |
| PolySkill2026.06 | 49.3 | 51.4 | 54.8 | 73.2 | 54.2 | 18.9 | 18.9 | 5.5 | 6.3 | 4.5 | 6.8 | 4 | 9.5 | 5.9 | |
| ASI2026.06 | 46.5 | 46.3 | 53.6 | 73.7 | 46.8 | 21.5 | 15.1 | 5.8 | 6.6 | 4.8 | 7.1 | 4.3 | 10 | 6.2 | |
| AWMonlineEvaluation Protocol=single-round, Task Initialization=starts without task-specific persistent state2026.06 | 46.4 | 48.1 | 49.3 | 68.9 | 40 | 45.3 | 5.7 | — | — | — | — | — | — | — | |
| SKILLMIGRATOR2026.06 | 45.7 | 45.5 | 52.7 | 72.6 | 46.7 | 19.3 | 16.7 | 4.8 | 5.6 | 4.5 | 6.7 | 3 | 9.4 | 5.4 | |
| SkillWeaver2026.06 | 43.6 | 39.3 | 48.2 | 71.2 | 50.3 | 17.2 | 16.3 | 5.7 | 6.5 | 4.7 | 7 | 4.2 | 9.8 | 6.1 | |
| ReAct2026.06 | 38.5 | 37.4 | 44 | 66 | 38.9 | 16.4 | 10.3 | 6.1 | 7 | 5 | 7.5 | 4.5 | 10.5 | 6.5 | |
| CodeActEvaluation Protocol=single-round, Task Initialization=starts without task-specific persistent state2026.06 | 38.3 | 39 | 44.5 | 55.7 | 31.2 | 33.9 | 10.2 | — | — | — | — | — | — | — | |
| SGDRModel=GPT-4.1, # Steps=4.82026.06 | 37.5 | 34.6 | 47.7 | 35.9 | 34.2 | 32.3 | — | — | — | — | — | — | — | — | |
| CERModel=GPT-4.1, # Steps=6.42026.06 | 33.9 | 31 | 38.4 | 31.1 | 37.1 | 28.6 | — | — | — | — | — | — | — | — | |
| ASIModel=GPT-4.1, # Steps=5.22026.06 | 33 | 29.6 | 41.4 | 33.8 | 29.7 | 29.4 | — | — | — | — | — | — | — | — | |
| VanillaModel=GPT-4.1, # Steps=6.02026.06 | 28.3 | 29 | 35.9 | 21.6 | 28 | 21.1 | — | — | — | — | — | — | — | — | |
| AWMModel=GPT-4.1, # Steps=5.92026.06 | 27.8 | 26.7 | 36.4 | 24.1 | 28.5 | 17.6 | — | — | — | — | — | — | — | — | |
| SGDRModel=QWEN3-4B, # Steps=5.62026.06 | 24.3 | 25.1 | 24.6 | 22.8 | 26.7 | 19.6 | — | — | — | — | — | — | — | — | |
| CERModel=QWEN3-4B, # Steps=6.52026.06 | 22.1 | 23.3 | 20.7 | 20.2 | 27.5 | 15.5 | — | — | — | — | — | — | — | — | |
| ASIModel=QWEN3-4B, # Steps=5.92026.06 | 20.8 | 22.5 | 22.6 | 19.2 | 22.3 | 14.2 | — | — | — | — | — | — | — | — | |
| VanillaModel=QWEN3-4B, # Steps=6.32026.06 | 16.5 | 21.9 | 14.8 | 12.7 | 16.6 | 13.5 | — | — | — | — | — | — | — | — | |
| AWMModel=QWEN3-4B, # Steps=5.72026.06 | 15.7 | 19.8 | 12.7 | 13.3 | 18.5 | 11.4 | — | — | — | — | — | — | — | — |