Mobile Navigation on GUIOdyssey updated evaluation set
79.2Step AccuracyAgentCPM-GUI-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| AgentCPM-GUI-8BModel Type=Open-source Models, Parameter Scale=8B2026.03 | 79.2 | |
| SecAgent-3BModel Type=Open-source Models, Parameter Scale=3B2026.03 | 74.3 | |
| UI-Venus-Navi-7BModel Type=Open-source Models, Parameter Scale=7B2026.03 | 71.1 | |
| OS-Atlas-7BModel Type=Open-source Models, Parameter Scale=7B2026.03 | 65.2 | |
| Qwen2.5-VL-7BModel Type=Open-source Models, Parameter Scale=7B2026.03 | 59.6 | |
| Qwen3-VL-8BModel Type=Open-source Models, Parameter Scale=8B2026.03 | 53.8 | |
| Qwen3-VL-4BModel Type=Open-source Models, Parameter Scale=4B2026.03 | 53.4 | |
| Qwen2.5-VL-3BModel Type=Open-source Models, Parameter Scale=3B2026.03 | 49.3 | |
| Claude-3-SonnetModel Type=Closed-source Models2026.03 | 46.5 | |
| GUI-R1-7BModel Type=Open-source Models, Parameter Scale=7B2026.03 | 41.9 | |
| UI-TARS-7BModel Type=Open-source Models, Parameter Scale=7B2026.03 | 37.5 | |
| GUI-R1-3BModel Type=Open-source Models, Parameter Scale=3B2026.03 | 34.6 | |
| UI-R1-3BModel Type=Open-source Models, Parameter Scale=3B2026.03 | 32.7 | |
| GPT-4oModel Type=Closed-source Models2026.03 | 29.3 |