Valley3: benchmark results
Provider: ByteDance. Released 2026-04-29.
Aggregate Rating 1611 ± 45, rank #910 of 10916 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MathVista | 75.2 | Accuracy (%) | 90.7 |
| HallusionBench | 58.1 | Overall Score | 86.9 |
| OpenVLM OCRBench | 87.2 | Score (normalized) | 83.1 |
| MMStar | 66.7 | Accuracy (%) | 79.7 |
| AI2D | 84.6 | Accuracy (%) | 76.1 |
| MM-Vet | 66.5 | Score | 74.3 |
| OpenCompass Multimodal - MMMU | 63.2 | Score (%) | 68.3 |
| OpenCompass Multimodal - MMBench | 83.1 | Score (%) | 66.3 |
| OpenCompass Multimodal - Average | 73.1 | Average Score | 64.4 |
Interactive version: theaggregate.ai/model?slug=valley3 · How It Works · Data refreshed daily, snapshot 2026-11-08.