Median Expert: benchmark results
Specialist benchmark reference assembled from studies with different expert groups and summary statistics; it does not estimate a population median. Provider: Human.
Aggregate Rating 1971 ± 23, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| GameWorld Computer-Use | 82.6 | Progress (%) | 100 |
| GameWorld Generalist | 82.6 | Progress (%) | 100 |
| LAB-Bench FigQA No Tools (Opus 4.6 System Card) | 77 | Accuracy (%) | 100 |
| Visual Aesthetic Benchmark (VAB) | 77.7 | Overall Top-1 ap@1 (imported) | 100 |
| MMMU Benchmark | 82.6 | Validation Score | 96.8 |
| ArchSIBench | 89.2 | Avg. (self-reported) | 95.2 |
| CodeElo | 1073 | Elo Rating | 91.2 |
| Chess Bench LLM | 1542 | Lichess-Classical-calibrated Glicko-2 rating (native signed | 88.6 |
| ALE-Bench | 1231 | Performance (Self-Refine x1) (imported) | 86.7 |
| Video-MMMU | 74.44 | Overall Score | 56.9 |
| PubMedQA | 78 | Accuracy (%) | 52.2 |
| LAB-Bench FigQA With Cropping (Opus 4.6 System Card) | 77 | Accuracy (%) | 50 |
Interactive version: theaggregate.ai/model?slug=median-expert · How It Works · Data refreshed daily, snapshot 2026-10-20.