Web Task Success on OnlineMind2Web
83.4Task Success RateFaraGen1.5 Solver (GPT-5.4)
Evaluation Results
| Method | Links | |
|---|---|---|
| FaraGen1.5 Solver (GPT-5.4)Size=—2026.06 | 83.4 | |
| Fara1.5-27BSize=27B2026.06 | 72.3 | |
| Yutori Navigator (n1)†Size=—2026.06 | 64.7 | |
| Fara1.5-9BSize=9B2026.06 | 63.4 | |
| OpenAI Operator†Size=—2026.06 | 58.3 | |
| GPT-5 SoMSize=—2026.06 | 57.7 | |
| Gemini 2.5 Computer Use†Size=—2026.06 | 57.3 | |
| Fara1.5-4BSize=4B2026.06 | 57.3 | |
| o3 SoMSize=—2026.06 | 55.4 | |
| GUI-Owl-1.5†Size=8B2026.06 | 48.6 | |
| MolmoWeb†Size=8B2026.06 | 35.3 | |
| Fara-7BSize=7B2026.06 | 34.1 | |
| Rule-based Step-RLTraining Paradigm=Self-evolving 4B, Ext. Expert Steps=02026.05 | 29.9 | |
| PRO-CUATraining Paradigm=Self-evolving 4B, Ext. Expert Steps=02026.05 | 28.8 | |
| PRO-CUA-8Training Paradigm=Self-evolving 8B, Ext. Expert Steps=02026.05 | 28.2 | |
| FBCTraining Paradigm=Self-evolving 8B, Ext. Expert Steps=02026.05 | 26.9 | |
| Rule-based Step-RLTraining Paradigm=Self-evolving 8B, Ext. Expert Steps=02026.05 | 26.2 | |
| WebSTAR-32BTraining Paradigm=External expert / closed data, Ext. Expert Steps=100K2026.05 | 23.8 | |
| FBCTraining Paradigm=Self-evolving 4B, Ext. Expert Steps=02026.05 | 23.7 | |
| GUI-Libra-4BTraining Paradigm=External expert / closed data, Ext. Expert Steps=81K2026.05 | 20 | |
| GUI-Libra-8BTraining Paradigm=External expert / closed data, Ext. Expert Steps=81K2026.05 | 19.3 | |
| WebSTAR-7BTraining Paradigm=External expert / closed data, Ext. Expert Steps=100K2026.05 | 17 | |
| Qwen3-VL-4B-InstructTraining Paradigm=Self-evolving 4B, Ext. Expert Steps=02026.05 | 16.7 | |
| UI-TARS-1.5-7BTraining Paradigm=External expert / closed data, Ext. Expert Steps=Closed data2026.05 | 14.6 | |
| Qwen3-VL-8B-InstructTraining Paradigm=Self-evolving 8B, Ext. Expert Steps=02026.05 | 12.2 |