SAIL-VL2-8B: benchmark results
Provider: ByteDance. Released 2025-09-11. Access: Open.
Aggregate Rating 1603 ± 36, rank #984 of 10916 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM OCRBench | 91.2 | Score (normalized) | 98 |
| MathVista | 76.2 | Accuracy (%) | 91.7 |
| AI2D | 87.8 | Accuracy (%) | 91.5 |
| MMStar | 70.7 | Accuracy (%) | 90.5 |
| MM-Vet | 70.7 | Score | 82.4 |
| VISTA-Bench - Multimodal Perception | 68.7 | Accuracy (%) on the 300 multimodal perception questions (glo | 77.6 |
| HallusionBench | 53.7 | Overall Score | 76.1 |
| OpenCompass Multimodal - MMBench | 84.5 | Score (%) | 76 |
| VISTA-Bench - Multimodal Reasoning | 54.3 | Accuracy (%) on the 300 multimodal reasoning questions (logi | 72.4 |
| OpenCompass Multimodal - Average | 74 | Average Score | 69.2 |
| VISTA-Bench (Pure Text) | 62 | Accuracy (%) over the same 1,500 questions given as ordinary | 63.8 |
| VISTA-Bench | 54 | Accuracy (%) over all 1,500 questions (300 perception, 300 r | 62.1 |
Interactive version: theaggregate.ai/model?slug=sail-vl2-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.