VL-Rethinker-72B: benchmark results
Provider: Other.
Aggregate Rating 1585 ± 48, rank #1187 of 10916 rated models, from 7 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| VLM-DeflectionBench - Realistic Accuracy | 47.2 | Accuracy (%) of knowledge-based visual question answering in | 66.7 |
| MME-Reasoning | 31.4 | Overall Accuracy (%) | 61.1 |
| VLM-DeflectionBench - Oracle Accuracy | 62.6 | Accuracy (%) of knowledge-based visual question answering in | 61.1 |
| MathVision | 44.93 | Overall Accuracy (%) | 58 |
| VLM-DeflectionBench - Oracle Hallucination Rate | 24.5 | Hallucination rate (%): share of answers that are wrong rath | 50 |
| VLM-DeflectionBench - Adversarial Hallucination Rate | 55.4 | Hallucination rate (%): share of answers that are wrong rath | 38.9 |
| VLM-DeflectionBench - Realistic Hallucination Rate | 40.9 | Hallucination rate (%): share of answers that are wrong rath | 27.8 |
Interactive version: theaggregate.ai/model?slug=vl-rethinker-72b · How It Works · Data refreshed daily, snapshot 2026-11-08.