Temporal Grounding on CoMET-Bench
19.3mIoUCoMET-Agent (Gemini 3 Flash)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CoMET-Agent (Gemini 3 Flash)#Size=-2026.06 | 19.3 | 19.5 | 19 | |
| CoMET-Agent (GPT-5)#Size=-2026.06 | 17.5 | 17.6 | 16.2 | |
| Gemini 3 Flash#Size=-2026.06 | 16.5 | 15.8 | 14.6 | |
| VideoARM#Size=-2026.06 | 13.5 | 12.8 | 13 | |
| Gemini 2.5 Pro#Size=-2026.06 | 10.8 | 9.9 | 10.1 | |
| LongVILA-R1#Size=7B2026.06 | 10.8 | 7.6 | 4.6 | |
| GPT-5#Size=-2026.06 | 9.7 | 9.5 | 10.1 | |
| CoMET-Agent (Qwen3-VL)#Size=8.6B2026.06 | 9.2 | 8.5 | 8.8 | |
| VideoMind#Size=7B2026.06 | 7.6 | 4.8 | 2.3 | |
| TimeLens#Size=7B2026.06 | 6.2 | 5 | 5.4 | |
| Qwen3-VL#Size=8.3B2026.06 | 5.1 | 3.9 | 3.4 | |
| TRACE#Size=7B2026.06 | 4 | 3.3 | 3.8 | |
| Qwen3-VL#Size=30B2026.06 | 3.3 | 3 | 3.1 | |
| DisTime#Size=7B2026.06 | 2.7 | 0.5 | 0.5 | |
| InternVL3.5#Size=38B2026.06 | 2.6 | 1.8 | 1.3 | |
| Eagle2.5#Size=8B2026.06 | 2.1 | 0.9 | 0.7 | |
| InternVL3.5#Size=8B2026.06 | 1.9 | 1.7 | 1.1 | |
| LITA#Size=13B2026.06 | 1.8 | 0.4 | 0.4 | |
| LLaVA-Video#Size=7B2026.06 | 0.7 | 0.3 | 0.4 | |
| LLaVA-OV1.5#Size=7B2026.06 | 0.3 | 0.2 | 0.2 | |
| MiMO-VL#Size=7B2026.06 | 0.3 | 0.1 | 0.1 |