VideoChat-R1: benchmark results
Provider: Other.
Aggregate Rating 1576 ± 49, rank #1317 of 10916 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EG-VQA - Evidence Grounding | 18.71 | Evidence-grounded F1 (%, EG-F1 at temporal IoU 0.3 and seman | 100 |
| EG-VQA - Descriptive | 13.24 | Strict accuracy (%) on descriptive questions (directly obser | 75 |
| EG-VQA - Relaxed Accuracy | 32.21 | Relaxed accuracy (%): fully correct answers count 1 and part | 75 |
| EG-VQA - Temporal | 9.99 | Strict accuracy (%) on temporal questions (event order and p | 75 |
| MuseBench - Cinematic Arts (Single-Select) | 35.87 | Cinematic-arts chance-adjusted accuracy (%): raw single-sele | 63 |
| MuseBench - Game Arts (Multi-Select) | 8.41 | Game-arts exact-match accuracy (%) on multi-select questions | 63 |
| MuseBench - Static Visual Arts (Multi-Select) | 12.1 | Static-visual-arts exact-match accuracy (%) on multi-select | 59.3 |
| MuseBench - Stage Performing Arts (Single-Select) | 33.04 | Stage-performing-arts chance-adjusted accuracy (%): raw sing | 55.6 |
| MuseBench - Static Visual Arts (Single-Select) | 29.05 | Static-visual-arts chance-adjusted accuracy (%): raw single- | 51.9 |
| EG-VQA | 16.87 | Strict accuracy (%): share of answers the judge rates fully | 50 |
| EG-VQA - Causal | 24.1 | Strict accuracy (%) on causal questions (why an event happen | 50 |
| MuseBench - Single-Select (Overall) | 26.49 | Overall chance-adjusted accuracy (%): raw single-select accu | 48.1 |
Interactive version: theaggregate.ai/model?slug=videochat-r1 · How It Works · Data refreshed daily, snapshot 2026-11-08.