Web Navigation on WebVoyager (test)
87Success RateOpenAI-CUA
Evaluation Results
| Method | Links | |
|---|---|---|
| OpenAI-CUA2026.02 | 87 | |
| UI-TARS-1.52026.02 | 84.8 | |
| Claude-3.72026.02 | 84.1 | |
| Holo2-30B-A3BParams.=30B2026.02 | 83 | |
| Holo2-4BParams.=4B2026.02 | 80.2 | |
| Holo2-8BParams.=8B2026.02 | 80.2 | |
| UI-Venus-1.5-30B-A3BParams.=30B2026.02 | 76 | |
| UI-Venus-1.5-8BParams.=8B2026.02 | 70.8 | |
| WebVoyager2026.02 | 59.1 | |
| UI-Venus-1.5-2BParams.=2B2026.02 | 56.4 | |
| GPT-4o2026.02 | 55.5 | |
| Qwen3-VL-30B-A3BParams.=30B, Evaluated by current paper=true2026.02 | 47.5 | |
| Qwen3-VL-8BParams.=8B, Evaluated by current paper=true2026.02 | 45.2 | |
| Qwen3-VL-2BParams.=2B, Evaluated by current paper=true2026.02 | 35.2 |