MiniCPM-V-4.6: benchmark results
Provider: OpenBMB. Released 2026-05-11. Access: Open.
Aggregate Rating 1442 ± 34, rank #1014 of 1635 rated models, from 55 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MathVista | 74.4 | Accuracy (%) | 89.3 |
| AA TAU-2 Bench | 87.72 | Accuracy (%) | 81.9 |
| Tau2-Bench Telecom | 87.72 | Success Rate (%) | 81.7 |
| MMStar | 65.7 | Accuracy (%) | 77.7 |
| HallusionBench | 53.9 | Overall Score | 76.9 |
| AI2D | 82.4 | Accuracy (%) | 63.1 |
| OpenVLM OCRBench | 80.6 | Score (normalized) | 60.9 |
| MM-Vet | 58.6 | Score | 55.8 |
| OpenCompass Multimodal - Average | 68.1 | Average Score | 43.3 |
| AA Omniscience - Software Engineering (SWE) - Dart | 14 | Accuracy (%) | 36.2 |
| OpenCompass Multimodal - MMMU | 51.9 | Score (%) | 35.6 |
| PureDocBench - Clean - Reading Order Edit Distance | 43.7 | Reading-order normalized edit distance (%) between the predi | 35.1 |
Interactive version: theaggregate.ai/model?slug=minicpm-v-4-6 · How It Works · Data refreshed daily, snapshot 2026-11-05.