Qwen 3 VL 8B (Thinking): benchmark results
Alibaba Qwen 3 VL 8B vision-language model evaluated with thinking enabled. Provider: Alibaba. Released 2025-10-15. Access: Open.
Aggregate Rating 1545 ± 5, rank #899 of 2168 rated models, from 670 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EXPLORE-Bench | 50.96 | Unified score (0-100): 0.25 x object coverage + 0.35 x attri | 100 |
| EXPLORE-Bench - Object Coverage | 62.7 | Object-level coverage (0-100): soft Sentence-BERT matching o | 100 |
| EXPLORE-Bench - Short Sequences | 52.08 | Unified score (0-100): 0.25 x object coverage + 0.35 x attri | 100 |
| FailBench - simplerenv | 79 | Balanced Accuracy (%) | 100 |
| VideoVIBE - Visual Hierarchy | 51.11 | Failure-diagnosis accuracy (%; questions of the Visual Hiera | 100 |
| EXPLORE-Bench - Attribute Score | 1.84 | Attribute-level score (0-5): a Qwen3-8B scorer rates the pre | 96.3 |
| EXPLORE-Bench - Medium Sequences | 50.56 | Unified score (0-100): 0.25 x object coverage + 0.35 x attri | 96.3 |
| EXPLORE-Bench - Relation Score | 2.8 | Relation-level score (0-5): a Qwen3-8B scorer rates the pred | 96.3 |
| DatBench - Chart Understanding | 73.2 | Accuracy (%) | 96.2 |
| DatBench - General VQA | 62.4 | Accuracy (%) | 96.2 |
| DatBench - Math & Logic | 43.3 | Accuracy (%) | 96.2 |
| OCR-Robust | 79.67 | OCR1.0 Clean (self-reported) | 94.1 |
Interactive version: theaggregate.ai/model?slug=qwen-3-vl-8b-thinking · How It Works · Data refreshed daily, snapshot 2026-10-20.