VST-7B-RL: benchmark results
Provider: Other.
Aggregate Rating 1612 ± 30, rank #905 of 10916 rated models, from 37 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CourtSI-Bench - Player-Zone | 53.6 | Exact-match accuracy (%) on the 82 player-to-court-zone rela | 87.5 |
| Metric-Bench | 38.11 | Mean relative accuracy (%; all questions; 1,340 template que | 85.7 |
| Metric-Bench - Depth | 55.43 | Mean relative accuracy (%; object-depth questions; 1,340 tem | 85.7 |
| Metric-Bench - Size | 40.76 | Mean relative accuracy (%; length, width, height, elevation | 85.7 |
| CourtSI-Bench - Player-Player | 64.9 | Exact-match accuracy (%) on the 393 player-to-player relatio | 81.8 |
| CourtSI-Bench | 40 | Overall score (%): the question-weighted mean of the 13 task | 79.5 |
| CourtSI-Bench - Player-Line | 61.8 | Exact-match accuracy (%) on the 495 player-to-court-line rel | 72.7 |
| CourtSI-Bench - Camera-Player | 75.4 | Exact-match accuracy (%) on the 248 camera-to-player relatio | 71.7 |
| CourtSI-Bench - Ball-Player | 59.6 | Exact-match accuracy (%) on the 297 ball-to-player relation | 68.2 |
| CourtSI-Bench - Object-Line Distance | 22.2 | Threshold mean relative accuracy (%), averaged over confiden | 68.2 |
| MultihopSpatial - 2-Hop Ego-Centric | 23.9 | Multiple-choice accuracy (%) on the 2-hop ego-centric questi | 65.3 |
| OVO-S-Bench - Egocentric Perception | 45.66 | Level 1 (instantaneous egocentric perception) multiple-choic | 60 |
Interactive version: theaggregate.ai/model?slug=vst-7b-rl · How It Works · Data refreshed daily, snapshot 2026-11-08.