Trajectory Completion Judgment on OGRBench Mobile
91AccuracyStainFlow
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| StainFlowVerifier=Gemini-3-Flash2026.06 | 91 | 91.3 | |
| OS-ThemisVerifier=Gemini-3-Flash2026.06 | 89.9 | 90.4 | |
| StainFlowVerifier=Qwen3.5-27B2026.06 | 89.4 | 90 | |
| StainFlowVerifier=Qwen3.5-9B2026.06 | 86.7 | 87.3 | |
| OS-ThemisVerifier=Qwen3.5-27B2026.06 | 85.6 | 86.2 | |
| StainFlowVerifier=GPT-52026.06 | 85.6 | 87.2 | |
| OS-ThemisVerifier=Qwen3-VL-8B2026.06 | 85.1 | 84.4 | |
| OS-ThemisVerifier=Qwen3.5-9B2026.06 | 85.1 | 84.6 | |
| OS-ThemisVerifier=GPT-52026.06 | 84 | 82.8 | |
| StainFlowVerifier=Qwen3-VL-8B2026.06 | 83 | 85.6 | |
| ZeroGUIVerifier=GPT-52026.06 | 81.4 | 83.1 | |
| ZeroGUIVerifier=Gemini-3-Flash2026.06 | 80.3 | 82.8 | |
| ZeroGUIVerifier=Qwen3.5-9B2026.06 | 77.1 | 79.2 | |
| ZeroGUIVerifier=Qwen3.5-27B2026.06 | 75.5 | 78.9 | |
| ZeroGUIVerifier=Qwen3-VL-8B2026.06 | 73.4 | 79.3 | |
| DigiRLVerifier=Gemini-3-Flash2026.06 | 72.9 | 71.8 | |
| DigiRLVerifier=Qwen3-VL-8B2026.06 | 71.3 | 70 | |
| DigiRLVerifier=Qwen3.5-27B2026.06 | 67.6 | 66.3 | |
| DigiRLVerifier=GPT-52026.06 | 64.9 | 62.9 | |
| DigiRLVerifier=Qwen3.5-9B2026.06 | 64.4 | 55.6 |