Browser Use on WebVoyager
90.6Success RateSoM Agent (GPT-5)
Evaluation Results
| Method | Links | |
|---|---|---|
| SoM Agent (GPT-5)Model Category=API only, Modality=w/ vision, # Steps=1002026.04 | 90.6 | |
| Gemini computer-use-previewModel Category=API only, Modality=w/ vision, # Steps=1002026.04 | 88.6 | |
| Axtree Agent (Gemini-3-flash)Model Category=API only, Modality=w/o vision, # Steps=1002026.04 | 85.6 | |
| SoM Agent (o3)Model Category=API only, Modality=w/ vision, # Steps=1002026.04 | 79.3 | |
| MolmoWeb-8BModel Category=Open weight, # Steps=1002026.04 | 78.2 | |
| MolmoWeb-4BModel Category=Open weight, # Steps=1002026.04 | 75.2 | |
| Axtree Agent (Gemini-3-flash)Model Category=API only, Modality=w/o vision, # Steps=302026.04 | 74.4 | |
| Fara-7BModel Category=Open weight, # Steps=1002026.04 | 73.5 | |
| OpenAI computer-use-previewModel Category=API only, Modality=w/ vision, # Steps=1002026.04 | 70.9 | |
| Axtree Agent (GPT-5)Model Category=API only, Modality=w/o vision, # Steps=302026.04 | 70.6 | |
| GLM-4.1V-9B-ThinkingModel Category=Open weight, # Steps=1002026.04 | 66.8 | |
| UI-TARS-1.5-7BModel Category=Open weight, # Steps=1002026.04 | 66.4 | |
| SoM Agent (GPT-4o)Model Category=API only, Modality=w/ vision, # Steps=1002026.04 | 65.1 | |
| Holo1-7BModel Category=Open weight, # Steps=302026.04 | 55.4 |