GUI agent success on WebVoyager (test)
90.6Success RateGPT-5 (SoM)
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5 (SoM)# Steps=1002026.05 | 90.6 | |
| Gemini computer-use-preview# Steps=1002026.05 | 88.6 | |
| Gemini-3-flash (Axtree)# Steps=1002026.05 | 85.6 | |
| o3 (SoM)# Steps=1002026.05 | 79.3 | |
| MolmoWeb-8B# Steps=100, # Tasks=>278.5K2026.05 | 78.2 | |
| MolmoWeb-4B# Steps=100, # Tasks=>278.5k2026.05 | 75.2 | |
| Gemini-3-flash (Axtree)# Steps=302026.05 | 74.4 | |
| Orchard-GUI-4B (SFT + RL)# Steps=30, # Tasks=2.6k2026.05 | 74.1 | |
| Fara-7B# Steps=100, # Tasks=>123.2k2026.05 | 73.5 | |
| OpenAI computer-use-preview# Steps=1002026.05 | 70.9 | |
| GPT-5 (Axtree)# Steps=302026.05 | 70.6 | |
| GLM-4.1V-9B-Thinking# Steps=1002026.05 | 66.8 | |
| UI-TARS-1.5-7B# Steps=1002026.05 | 66.4 | |
| GPT-4o (SoM)# Steps=1002026.05 | 65.1 | |
| Qwen3-VL-235B-A22B-Thinking# Steps=302026.05 | 63.1 | |
| Orchard-GUI-4B-SFT# Steps=30, # Tasks=0.4k2026.05 | 60.2 | |
| Holo1-7B# Steps=30, # Tasks=>15.6k2026.05 | 55.4 | |
| Qwen3-VL-4B-Thinking# Steps=302026.05 | 49 |