MLVU mean average (MLVU (M-Avg))
We show this table for reference; we do not rank on it.
A multi-task video understanding benchmark averaged across MLVU categories.
Benchmark score on MLVU (M-Avg) — November 13, 2026
We compile the MLVU (M-Avg) rows from provider self-reports and secondary reports. Qwen3.8 Max leads the table at 90.8%, followed by Qwen3.7 Plus (87.4%) and Qwen3.5 397B (86.7%). We do not use these results to rank models overall.
Qwen3.8 Max
Alibaba
Reported by AlibabaQwen3.7 Plus
Alibaba
Reported by AlibabaQwen3.5 397B
Alibaba
Reported by Alibaba10 modelsMultimodal & GroundedCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (10 models)
None of the 10 results on this page has a documented run by the benchmark or an independent evaluator: 6 reported by the labs that built the models and 4 reported in another publisher’s document.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Qwen3.8 MaxAlibabaReported by Alibaba | 90.8% | Reported by Alibaba | Not reported | Open |
| 2 | Qwen3.7 PlusAlibabaReported by Alibaba | 87.4% | Reported by Alibaba | Not reported | Closed |
| 3 | Qwen3.5 397BAlibabaReported by Alibaba | 86.7% | Reported by Alibaba | Not reported | Open |
| 4 | Qwen3.6-27BAlibabaReported by Alibaba | 86.6% | Reported by Alibaba | Not reported | Open |
| 5 | Qwen3.6-35B-A3BAlibabaReported by Alibaba | 86.2% | Reported by Alibaba | Not reported | Open |
| 6 | GPT-5.2OpenAIAnother lab’s comparison table | 85.6% | Another lab’s comparison table | Not reported | Closed |
| 7 | Kimi K2.5Moonshot AIAnother lab’s comparison table | 85.0% | Another lab’s comparison table | Not reported | Open |
| 8 | Gemini 3 ProGoogleAnother lab’s comparison table | 83.0% | Another lab’s comparison table | Not reported | Closed |
| 9 | Claude Opus 4.5AnthropicAnother lab’s comparison table | 81.7% | Another lab’s comparison table | Not reported | Closed |
| 10 | LFM3-VL-2BLiquidAIReported by LiquidAI | 71.3% | Reported by LiquidAI | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Among the reported MLVU (M-Avg) rows, Qwen3.8 Max is first at 90.8%. The third row is 4.1 points behind. The broader top-10 range is 19.5 points, so the table still separates the published systems.
10 models have been evaluated on MLVU (M-Avg). The benchmark falls in the Multimodal & Grounded category. MLVU (M-Avg) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About MLVU (M-Avg)
Year
2026
Tasks
General video understanding
Format
Video QA and understanding
Difficulty
Broad multimodal video reasoning
MLVU captures general-purpose video understanding rather than a single narrow skill. BenchLM tracks the mean-average summary row so provider comparison tables can be compared directly.
Freshness and provenance
Version
MLVU (M-Avg) 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 MLVU (M-Avg) measure?
A multi-task video understanding benchmark averaged across MLVU categories.
Which model scores highest on MLVU (M-Avg)?
Qwen3.8 Max by Alibaba currently leads with a score of 90.8% on MLVU (M-Avg).
How many models are evaluated on MLVU (M-Avg)?
10 AI models have published results on MLVU (M-Avg) in the BenchLM catalog.
Compare top models on MLVU (M-Avg)
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.