Qwen 3 VL 4B Instruct — benchmark results
Alibaba's Apache-2.0 4B dense vision-language model (October 2025) with 256K context, the compact tier of the Qwen3-VL line. Provider: Alibaba. Released 2025-07-01. Access: Open.
Unified ELO 1467 ± 13, rank #921 of 1776 rated models, from 104 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LatamBoard - ASSIN2 RTE | 94.08 | Score (%) | 93.9 |
| LatamBoard - FaQuAD NLI | 81.37 | Score (%) | 93.9 |
| LatamBoard - Portuguese Score | 89.58 | Score (%) | 93.9 |
| LatamBoard - BLUEX | 67.59 | Score (%) | 90.9 |
| LatamBoard - ENEM Challenge | 76.35 | Score (%) | 90.9 |
| LatamBoard - OAB Exams | 53.39 | Score (%) | 90.9 |
| LatamBoard - Spanish WNLI | 76.06 | Score (%) | 83.3 |
| UGI - Willingness (W/10) | 7.8 | W/10 Score | 81.1 |
| LLM Stats (ODinW) | 48.2 | Score (%) | 73.3 |
| LatamBoard - Spanish XNLI | 47.35 | Score (%) | 66.7 |
| TriViewBench | 52 | Overall (self-reported) | 66.7 |
| LatamBoard - Spanish TeleIA | 66.67 | Score (%) | 62.1 |
Interactive version: theaggregate.ai/model?slug=qwen-3-vl-4b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.