Mobile Agent Task Completion on Mobile-Eval-RAG
79.6Completion RateMobile-Agent-RAG
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Mobile-Agent-RAGBackbone MLLM=Claude-3.5-Sonnet2025.11 | 79.6 | 92.7 | 95.1 | 18.7 | 4.26 | 84 | |
| Mobile-Agent-RAGBackbone MLLM=GPT-4o2025.11 | 79.1 | 88.4 | 97.8 | 18.4 | 4.3 | 84 | |
| Mobile-Agent-E + EvoBackbone MLLM=Claude-3.5-Sonnet2025.11 | 76.7 | 91.3 | 95.6 | 17.1 | 4.49 | 72 | |
| Mobile-Agent-RAGBackbone MLLM=Gemini-1.5-Pro2025.11 | 75.7 | 90.1 | 94.7 | 18.8 | 4.03 | 76 | |
| Mobile-Agent-E + EvoBackbone MLLM=GPT-4o2025.11 | 73.3 | 85.6 | 92.4 | 17.8 | 4.12 | 76 | |
| Mobile-Agent-EBackbone MLLM=Claude-3.5-Sonnet2025.11 | 72.3 | 91 | 94.8 | 17.3 | 4.18 | 60 | |
| Mobile-Agent-EBackbone MLLM=GPT-4o2025.11 | 67 | 81.5 | 90.2 | 18.6 | 3.6 | 60 | |
| Mobile-Agent-E + EvoBackbone MLLM=Gemini-1.5-Pro2025.11 | 61.2 | 77.2 | 91 | 21.8 | 2.81 | 56 | |
| Mobile-Agent-EBackbone MLLM=Gemini-1.5-Pro2025.11 | 58.3 | 74.1 | 89.3 | 22.4 | 2.6 | 48 |