Visual Question Answering on RoboFine-Bench (test)
71Overall AccuracyRoboFine-VLM
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RoboFine-VLM2026.05 | 71 | 85.2 | 63.5 | 72.5 | 73.6 | 67.3 | 56.7 | 81.8 | 57.7 | 66.5 | 85.7 | |
| Gemini-3.1-Pro2026.05 | 62.1 | 83.6 | 67.1 | 68.8 | 72.9 | 52.6 | 52.1 | 63.6 | 23.1 | 67.6 | 50 | |
| GPT-5.42026.05 | 61 | 85.1 | 60 | 58.8 | 66.4 | 61.5 | 50.7 | 63.6 | 50 | 65.4 | 28.6 | |
| Doubao-Seed-2.0-Pro2026.05 | 54.9 | 60.7 | 55.3 | 61.3 | 61.4 | 50 | 45.1 | 72.7 | 42.3 | 61.6 | 50 | |
| Qwen3.5-Plus2026.05 | 52.6 | 70.5 | 47.1 | 62.5 | 55 | 45.5 | 47.4 | 72.7 | 26.9 | 58.4 | 42.9 | |
| Qwen3-VL-Plus2026.05 | 50.4 | 68.9 | 51.8 | 55 | 62.1 | 43 | 43.7 | 63.6 | 50 | 46 | 50 |