Multimodal Multi-disciplinary Video Understanding (MMVU)
We show this table for reference; we do not rank on it.
A benchmark for evaluating multimodal models on video understanding tasks across multiple disciplines, emphasizing temporal reasoning and comprehension over video content.
Benchmark score on MMVU — November 21, 2026
We compile the MMVU rows from provider self-reports and secondary reports. Qwen3.8 Max leads the table at 82.4%, followed by GLM-5.3-Flash (80.5%) and Kimi K2.5 (80.4%). We do not use these results to rank models overall.
Qwen3.8 Max
Alibaba
Reported by AlibabaGLM-5.3-Flash
Z.AI
Reported by Z.AIKimi K2.5
Moonshot AI
Carried from an upstream page8 modelsMultimodal & GroundedCurrentDisplay onlyUpdated November 21, 2026
Benchmark score table (8 models)
None of the 8 results on this page has a documented run by the benchmark or an independent evaluator: 4 reported by the labs that built the models and 4 carried from an upstream page, where who ran them is not recorded.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Qwen3.8 MaxAlibabaReported by Alibaba | 82.4% | Reported by Alibaba | Not reported | Open |
| 2 | GLM-5.3-FlashZ.AIReported by Z.AI | 80.5% | Reported by Z.AI | Not reported | Open |
| 3 | Kimi K2.5Moonshot AICarried from an upstream page | 80.4% | Carried from an upstream page | Not reported | Open |
| 4 | dots3-note PreviewDots StudioReported by Dots Studio | 79.9% | Reported by Dots Studio | Not reported | Open |
| 5 | Qwen3.5-122B-A10BAlibabaCarried from an upstream page | 74.7% | Carried from an upstream page | Not reported | Open |
| 6 | Qwen3.5-27BAlibabaCarried from an upstream page | 73.3% | Carried from an upstream page | Not reported | Open |
| 7 | Qwen3.5-35B-A3BAlibabaCarried from an upstream page | 72.3% | Carried from an upstream page | Not reported | Open |
| 8 | LFM3-VL-2BLiquidAIReported by LiquidAI | 52.9% | Reported by LiquidAI | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Among the reported MMVU rows, Qwen3.8 Max is first at 82.4%. The third row is 2.0 points behind. The broader top-10 range is 29.5 points, so the table still separates the published systems.
8 models have been evaluated on MMVU. The benchmark falls in the Multimodal & Grounded category. MMVU is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About MMVU
Year
2026
Tasks
Video understanding
Format
Video reasoning benchmark
Difficulty
Multi-disciplinary multimodal video reasoning
MMVU is a useful video-understanding benchmark for BenchLM because it appears in frontier provider tables and complements Video-MME with a more discipline-oriented video reasoning slice.
Freshness and provenance
Version
MMVU 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does MMVU measure?
A benchmark for evaluating multimodal models on video understanding tasks across multiple disciplines, emphasizing temporal reasoning and comprehension over video content.
Which model scores highest on MMVU?
Qwen3.8 Max by Alibaba currently leads with a score of 82.4% on MMVU.
How many models are evaluated on MMVU?
8 AI models have published results on MMVU in the BenchLM catalog.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.