Video Summarization on Video-MME 900 videos
85.2Overall AccuracyGemini 2.5 Pro
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Gemini 2.5 Proaudio transcription=w/2026.01 | 85.2 | 0.5 | |
| Gemini 2.5 Proaudio transcription=w/o2026.01 | 84.7 | — | |
| Gemini 1.5 Proaudio transcription=w/2026.01 | 81.3 | 6.3 | |
| Qwen2-VLaudio transcription=w/2026.01 | 77.8 | 6.6 | |
| GPT-4oaudio transcription=w/2026.01 | 77.2 | 8.2 | |
| LLaVA-Videoaudio transcription=w/2026.01 | 76.9 | 0.9 | |
| LLaVA-Videoaudio transcription=w/o2026.01 | 76 | — | |
| Gemini 1.5 Proaudio transcription=w/o2026.01 | 75 | — | |
| Gemini 1.5 Flashaudio transcription=w/2026.01 | 75 | 2.4 | |
| Oryx-1.5audio transcription=w/2026.01 | 74.9 | 7.6 | |
| InternVL2.5audio transcription=w/2026.01 | 74 | 6.4 | |
| Gemini 1.5 Flashaudio transcription=w/o2026.01 | 72.6 | — | |
| Ariaaudio transcription=w/2026.01 | 72.1 | 1.8 | |
| LinVTaudio transcription=w/2026.01 | 71.7 | 6.1 | |
| TPOaudio transcription=w/2026.01 | 71.5 | 5.3 | |
| Qwen2-VLaudio transcription=w/o2026.01 | 71.2 | — | |
| Ariaaudio transcription=w/o2026.01 | 70.3 | — | |
| GPT-4oaudio transcription=w/o2026.01 | 69 | — | |
| InternVL2.5audio transcription=w/o2026.01 | 67.6 | — | |
| Oryx-1.5audio transcription=w/o2026.01 | 67.3 | — | |
| TPOaudio transcription=w/o2026.01 | 66.2 | — | |
| LinVTaudio transcription=w/o2026.01 | 65.6 | — |