Video Evidence Grounding on EG-VQA (test)
70.02F1@0.1Human Results*
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Human Results*2026.06 | 70.02 | 56.26 | 21.19 | 8.43 | 34.79 | 6.56 | 3.87 | |
| EG-ReasonerBase Model=Qwen2.5-VL-7B-Instruct2026.06 | 65.97 | 48.02 | 27.27 | 10.3 | 28.79 | 4.95 | 3.12 | |
| VideoChat-R1-7BModel Category=Open-sourced Video-LLMs2026.06 | 48.24 | 34.5 | 18.43 | 6.84 | 18.71 | 2.7 | 1.56 | |
| Time-R1-7BModel Category=Open-sourced Video-LLMs2026.06 | 46.46 | 32.03 | 15.64 | 5.41 | 16.46 | 2.75 | 1.4 | |
| VideoChat-R1-Thinking-7BModel Category=Open-sourced Video-LLMs2026.06 | 40.9 | 26.9 | 13.69 | 4.77 | 14.51 | 1.82 | 0.94 | |
| GPT-4oModel Category=Commercial Large Multimodal Models2026.06 | 35.85 | 17.85 | 5.67 | 2.95 | 8.49 | 1.18 | 0.33 | |
| TW-GRPOModel Category=Open-sourced Video-LLMs2026.06 | 33.3 | 19.48 | 8.97 | 2.94 | 9.65 | 1.07 | 0.58 | |
| VideoChat-R1.5-7BModel Category=Open-sourced Video-LLMs2026.06 | 33.26 | 21.44 | 10.88 | 3.58 | 11.86 | 2.17 | 1.22 | |
| Qwen2.5-VL-7BModel Category=Open-sourced Video-LLMs2026.06 | 29.36 | 15.83 | 7.7 | 2.55 | 5.11 | 0.6 | 0.35 | |
| Gemini-2.5-FlashModel Category=Commercial Large Multimodal Models2026.06 | 28.87 | 19.81 | 12.56 | 6.88 | 12.64 | 1.48 | 0.86 | |
| InternVL-3-8BModel Category=Open-sourced Video-LLMs2026.06 | 20.89 | 11.02 | 4.68 | 1.34 | 8.18 | 1.15 | 0.66 |