Robot Video Diagnosis on RoboGazeBench (test)
75.5Desc. F1Human ceiling
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Human ceiling2026.06 | 75.5 | 71 | 47.1 | 94 | |
| Gemini 3.1 ProFamily=Prop., Prompting Strategy=ROBOGAZE (RG)2026.06 | 67.9 | 64 | 43.7 | 92 | |
| GPT-5.5Family=Prop., Prompting Strategy=ROBOGAZE (RG)2026.06 | 65.1 | 63 | 41.1 | 90 | |
| Claude Sonnet 4.6Family=Prop., Prompting Strategy=ROBOGAZE (RG)2026.06 | 60.3 | 61 | 36.8 | 89 | |
| Gemini 3.1 FlashFamily=Prop., Prompting Strategy=ROBOGAZE (RG)2026.06 | 57.8 | 61 | 35 | 87 | |
| Gemma4-31BFamily=Open, Prompting Strategy=ROBOGAZE (RG)2026.06 | 56.6 | 66 | 34.5 | 86 | |
| Qwen3.6-35BFamily=Open, Prompting Strategy=ROBOGAZE (RG)2026.06 | 53.1 | 60 | 31.6 | 84 | |
| InternVL3.5-38BFamily=Open, Prompting Strategy=ROBOGAZE (RG)2026.06 | 50.1 | 58 | 28.9 | 81 | |
| LLaVA-OV-2-8BFamily=Open, Prompting Strategy=ROBOGAZE (RG)2026.06 | 47.9 | 56 | 26.9 | 82 | |
| Gemini 3.1 ProFamily=Prop., Prompting Strategy=chain-of-thought (CoT)2026.06 | 32.4 | 43 | 13.9 | 21 | |
| GPT-5.5Family=Prop., Prompting Strategy=chain-of-thought (CoT)2026.06 | 30.5 | 41 | 12.5 | 18 | |
| Claude Sonnet 4.6Family=Prop., Prompting Strategy=chain-of-thought (CoT)2026.06 | 27.6 | 40 | 11 | 20 | |
| Gemini 3.1 FlashFamily=Prop., Prompting Strategy=chain-of-thought (CoT)2026.06 | 26.1 | 39 | 10.2 | 16 | |
| Gemini 3.1 ProFamily=Prop., Prompting Strategy=vanilla (V)2026.06 | 25.4 | 39 | 9.8 | 13 | |
| GPT-5.5Family=Prop., Prompting Strategy=vanilla (V)2026.06 | 23.9 | 37 | 8.8 | 9 | |
| Gemma4-31BFamily=Open, Prompting Strategy=chain-of-thought (CoT)2026.06 | 23.4 | 37 | 8.7 | 12 | |
| Qwen3.6-35BFamily=Open, Prompting Strategy=chain-of-thought (CoT)2026.06 | 22.3 | 36 | 8.2 | 23 | |
| Claude Sonnet 4.6Family=Prop., Prompting Strategy=vanilla (V)2026.06 | 21.4 | 36 | 7.7 | 11 | |
| InternVL3.5-38BFamily=Open, Prompting Strategy=chain-of-thought (CoT)2026.06 | 21 | 35 | 7.4 | 13 | |
| Gemini 3.1 FlashFamily=Prop., Prompting Strategy=vanilla (V)2026.06 | 20.2 | 35 | 7 | 8 | |
| LLaVA-OV-2-8BFamily=Open, Prompting Strategy=chain-of-thought (CoT)2026.06 | 19.5 | 33 | 6.5 | 25 | |
| Gemma4-31BFamily=Open, Prompting Strategy=vanilla (V)2026.06 | 17.6 | 33 | 5.7 | 3 | |
| Qwen3.6-35BFamily=Open, Prompting Strategy=vanilla (V)2026.06 | 16.7 | 32 | 5.4 | 16 | |
| InternVL3.5-38BFamily=Open, Prompting Strategy=vanilla (V)2026.06 | 15.5 | 31 | 4.8 | 4 | |
| LLaVA-OV-2-8BFamily=Open, Prompting Strategy=vanilla (V)2026.06 | 14.2 | 30 | 4.3 | 18 |