Cross-video difference understanding on DELTAVID-Bench 500 diagnostic questions (test)
78MCQ ScoreGPT-5
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-52026.06 | 78 | 55.43 | 77.78 | 70 | |
| Qwen3-VL-235BParameters=235B2026.06 | 72 | 58.86 | 67.11 | 65.2 | |
| DELTAVID-Qwen3-VL-8BBase Model=Qwen3-VL-8B, Training=DELTAVID2026.06 | 69 | 64 | 69.33 | 67.4 | |
| Video-R12026.06 | 68 | 30.29 | 62.22 | 52.2 | |
| Qwen3-VL-8BParameters=8B2026.06 | 65 | 40.57 | 69.78 | 58.6 |