SAIL-VL2-2B: benchmark results
Provider: ByteDance. Released 2025-09-11. Access: Open.
Aggregate Rating 1518 ± 39, rank #2094 of 10916 rated models, from 23 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM OCRBench | 90.6 | Score (normalized) | 95.5 |
| MathVista | 71.5 | Accuracy (%) | 82.9 |
| MMStar | 64 | Accuracy (%) | 72.8 |
| MM-Vet | 65.6 | Score | 71.4 |
| HallusionBench | 51.1 | Overall Score | 70.8 |
| AI2D | 83.1 | Accuracy (%) | 68.6 |
| VISTA-Bench - Multimodal Perception | 65.3 | Accuracy (%) on the 300 multimodal perception questions (glo | 53.4 |
| OpenCompass Multimodal - Average | 69.4 | Average Score | 48.6 |
| OpenCompass Multimodal - MMBench | 80.1 | Score (%) | 44.7 |
| VISTA-Bench - Multimodal Knowledge | 32.8 | Accuracy (%) on the 400 multimodal knowledge questions (STEM | 41.4 |
| VISTA-Bench - Multimodal Reasoning | 47.3 | Accuracy (%) on the 300 multimodal reasoning questions (logi | 41.4 |
| OddGridBench - Size | 2.5 | Accuracy (%) on the 200 images in which the odd item differs | 36.1 |
Interactive version: theaggregate.ai/model?slug=sail-vl2-2b · How It Works · Data refreshed daily, snapshot 2026-11-08.