video-SALMONN-2+ 7B: benchmark results
Provider: Other. Access: Open.
Aggregate Rating 1539 ± 28, rank #1841 of 10916 rated models, from 38 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OmniPro - Event Alert | 37.2 | Probe-mode accuracy (%): for each ground-truth trigger the m | 100 |
| OmniPro - Target Grounding | 18.1 | Probe-mode accuracy (%): for each ground-truth trigger the m | 100 |
| OmniPro - Snapshot Counting | 24.7 | Probe-mode accuracy (%): for each ground-truth trigger the m | 85.7 |
| OmniPro | 22.1 | Probe-mode accuracy (%): for each ground-truth trigger the m | 71.4 |
| OmniPro - Condition Alert | 17.6 | Probe-mode accuracy (%): for each ground-truth trigger the m | 71.4 |
| OmniPro - Cumulative Counting | 11.5 | Probe-mode accuracy (%): for each ground-truth trigger the m | 71.4 |
| OmniPro - State Monitoring | 12.3 | Probe-mode accuracy (%): for each ground-truth trigger the m | 57.1 |
| LVOmniBench - Perception | 36.3 | Accuracy (%) on the perception questions (object attributes, | 50 |
| MTAVG-Bench - Audio-Visual Lip Synchronization | 54.48 | Score (%) on the questions diagnosing alignment of lip motio | 45.5 |
| MTAVG-Bench - Scene Consistency | 41.46 | Score (%) on the questions diagnosing coherence of location, | 45.5 |
| OmniPro - Deduplicated Counting | 20.3 | Probe-mode accuracy (%): for each ground-truth trigger the m | 42.9 |
| OmniPro - Event Narration | 41.3 | Probe-mode accuracy (%): for each ground-truth trigger the m | 42.9 |
Interactive version: theaggregate.ai/model?slug=video-salmonn-2-plus-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.