Temporal Grounding on OphVL In-domain (test)
87.85R@0.3Gemini 3-Pro-Preview
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini 3-Pro-PreviewModel Class=Proprietary LMM2026.02 | 87.85 | 73.06 | 57.57 | 72.43 | |
| MedScope-7B-RLModel Class=Reasoning Agent, Training=RL2026.02 | 81.92 | 68.87 | 54.35 | 64.42 | |
| MedScope-7B-SFTModel Class=Reasoning Agent, Training=SFT2026.02 | 67.19 | 54.43 | 43.72 | 49.28 | |
| Video-R1-7BModel Class=Reasoning LMM2026.02 | 63.94 | 22.89 | 7.04 | 33.87 | |
| Gemini-2.5-FlashModel Class=Proprietary LMM2026.02 | 59.33 | 48.77 | 13.5 | 68.31 | |
| LongVT-7B-RFTModel Class=Reasoning Agent2026.02 | 48.73 | 32.32 | 23.52 | 39.56 | |
| ReWatch-R1-7BModel Class=Reasoning Agent2026.02 | 45.85 | 30.56 | 26.52 | 24.77 | |
| Qwen2.5-VL-InstructModel Class=Open-Source LMM, Parameters=7B2026.02 | 43.45 | 41.16 | 21.94 | 30.18 | |
| InternVL3-8BModel Class=Open-Source LMM2026.02 | 42.15 | 36.65 | 22.57 | 29.01 | |
| GPT-4oModel Class=Proprietary LMM2026.02 | 41.73 | 34.33 | 29.23 | 35.75 | |
| VideoLLaMA3-7BModel Class=Open-Source LMM2026.02 | 37.36 | 25.01 | 14.39 | 23.1 | |
| VideoChat-R1-7BModel Class=Reasoning LMM2026.02 | 30.58 | 21.41 | 9.17 | 30.93 | |
| Video-RFTModel Class=Reasoning LMM2026.02 | 26.41 | 19.37 | 7.04 | 27.06 |