Multi-modal Cloze Test on Omni-Cloze (test)
31.5Visual AccuracyGemini-2.5-Flash
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini-2.5-Flash2026.06 | 31.5 | 18.4 | 39.1 | 27.9 | |
| OmniCaptioner-IF-7BParameters=7B2026.06 | 23.86 | 24.23 | 32.3 | 25.17 | |
| OmniCaptioner-IF-3BParameters=3B2026.06 | 21.27 | 21.81 | 28.94 | 22.53 | |
| Qwen2.5-Omni-7BParameters=7B2026.06 | 10.4 | 12.9 | 18.9 | 12.9 | |
| VideoLLaMA-2-7BParameters=7B2026.06 | 5.7 | 2.6 | 7.3 | 4.8 | |
| video-SALMONN-13BParameters=13B2026.06 | 2.6 | 1.7 | 4 | 2.5 |