Qwen 3.5 4B Instruct: benchmark results
Provider: Alibaba. Released 2026-03-02.
Aggregate Rating 1557 ± 36, rank #1606 of 10916 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| VideoJev Evaluation - LVBench test without subtitles | 60 | Accuracy (%) | 47.4 |
| VideoJev Evaluation - Video-MME test without subtitles | 70.6 | Accuracy (%) | 22.2 |
| Video-MME-v2 - Level 3 Complex Reasoning (w/o subs) | 8.5 | Level 3 Score w/o sub (%) | 14.6 |
| Video-MME-v2 | 24.1 | Avg Accuracy w/o sub (%) | 10.4 |
| Video-MME-v2 - Level 2 Temporal Understanding (w/o subs) | 9.04 | Level 2 Score w/o sub (%) | 10.4 |
| Video-MME-v2 - Non-Lin Score (w/o subs) | 9.5 | Non-Lin Score w/o sub (%) | 10.4 |
| Video-MME-v2 - Level 1 Retrieval & Aggregation (w/o subs) | 11.88 | Level 1 Score w/o sub (%) | 6.2 |
| VideoJev Evaluation - LongVideoBench validation without subtitles | 61.2 | Accuracy (%) | 0 |
| VideoJev Evaluation - MLVU development MCQ | 73.3 | Accuracy (%) | 0 |
Interactive version: theaggregate.ai/model?slug=qwen-3-5-4b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.