Web Navigation on Multimodal-Mind2Web Cross-Domain
57.4Element AccuracySkill-CMIB
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Skill-CMIBLearning Paradigm=Skill Based2026.05 | 57.4 | 69.5 | 42.9 | |
| AgentTrek-7BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | 56 | 87.5 | 52.6 | |
| Explorer-7BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | 55.7 | 90.4 | 53 | |
| Explorer-7BLearning Paradigm=Supervised Fine-Tuning2026.05 | 50.9 | 88.9 | 48.1 | |
| Explorer-4BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | 49.8 | 88.8 | 47.2 | |
| Explorer-4BLearning Paradigm=Supervised Fine-Tuning2026.05 | 46.9 | 87.7 | 44.6 | |
| Qwen2.5-VL-7B-Inst.Learning Paradigm=Skill Based2026.05 | 46.8 | 66.3 | 32.4 | |
| SeeActLearning Paradigm=In-Context Learning2026.05 | 42.4 | 69.3 | 36.8 | |
| Scribe-Agent-L.-32BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | 39.4 | 54.7 | 37.3 | |
| GPT-4Learning Paradigm=In-Context Learning2026.05 | 35.4 | 61.9 | 29.7 | |
| GPT-3.5Learning Paradigm=In-Context Learning2026.05 | 25.2 | 57.9 | 24.1 | |
| ClassificationLearning Paradigm=Supervised Fine-Tuning2026.05 | 24.5 | — | — | |
| SeeClick-9.6BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | 22.1 | 84.1 | 20.2 | |
| MiniCPM-GUI-3.1BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | 17.9 | 74.5 | 14.6 | |
| GenerationLearning Paradigm=Supervised Fine-Tuning2026.05 | 14.2 | 44.7 | 11.9 | |
| EDGE-9.6BLearning Paradigm=Data Synthesis+Supervised Fine-Tuning2026.05 | — | — | 22.4 |