One-to-One Temporal Grounding on QVHighlights TimeLens-Bench
84.1R1@0.3Gemini-2.5-Pro
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini-2.5-ProModel Category=Proprietary Models2026.06 | 84.1 | 75.9 | 61.1 | 70.4 | |
| OMTG-4BModel Category=Open-Source Models, Training Stage=RL2026.06 | 80.9 | 70.4 | 53.5 | 62.2 | |
| Qwen3-VL-4BModel Category=Open-Source Models2026.06 | 77.6 | 66 | 50.9 | 61.7 | |
| OMTG-4BModel Category=Open-Source Models, Training Stage=SFT2026.06 | 76.5 | 65.4 | 50.8 | 61.5 | |
| Qwen3-VL-8BModel Category=Open-Source Models2026.06 | 74.2 | 64.6 | 49.3 | 59.4 | |
| TimeLens-7BModel Category=Open-Source Models2026.06 | 74.1 | 62.7 | 43.1 | 56 | |
| GPT-oModel Category=Proprietary Models2026.06 | 72.4 | 60.4 | 46.4 | 56.8 | |
| GPT-4oModel Category=Proprietary Models2026.06 | 69 | 54.8 | 38.5 | 52.1 | |
| Time-R1-7BModel Category=Open-Source Models2026.06 | 65.8 | 51.5 | 36.1 | 49.2 | |
| VideoChat-Flash-7BModel Category=Open-Source Models2026.06 | 45.2 | 30.6 | 16.7 | 32.7 | |
| Qwen2.5-VL-7BModel Category=Open-Source Models2026.06 | 41.5 | 27.8 | 15.2 | 31.6 | |
| VideoChat-R1-7BModel Category=Open-Source Models2026.06 | 29.3 | 19.1 | 9.4 | 21.5 |