multimodal benchmark
MVBench Leaderboard
MVBench is a multimodal video understanding benchmark covering 20 video tasks that require temporal understanding beyond single-frame analysis.
Updated Sep 5, 2026
Models18
Model coverage18
MetricScore
EvidenceB
MVBench Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelZA | Score77.80% | Percentile100.00% | Participants18 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score76.60% | Percentile94.12% | Participants18 | EvidenceC | Evaluated |
| Rank03 | ModelAC | Score75.50% | Percentile88.24% | Participants18 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score74.80% | Percentile82.35% | Participants18 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score74.60% | Percentile76.47% | Participants18 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score74.60% | Percentile70.59% | Participants18 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score73.60% | Percentile64.71% | Participants18 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score73.20% | Percentile58.82% | Participants18 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score72.80% | Percentile52.94% | Participants18 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score72.30% | Percentile47.06% | Participants18 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score72.00% | Percentile41.18% | Participants18 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score70.40% | Percentile35.29% | Participants18 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score70.30% | Percentile29.41% | Participants18 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score69.60% | Percentile23.53% | Participants18 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score69.30% | Percentile17.65% | Participants18 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score69.00% | Percentile11.76% | Participants18 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score68.90% | Percentile5.88% | Participants18 | EvidenceC | Evaluated |
| Rank18 | ModelAC | Score68.70% | Percentile0.00% | Participants18 | EvidenceC | Evaluated |
MVBench Highlights
The leading models and scores on this benchmark.
MVBench Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for MVBench
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is MVBench?
What MVBench measures and how its scores work.
MVBench is a benchmark for multimodal video understanding with tasks spanning perception and cognition, including action recognition, temporal reasoning, spatial reasoning, object interaction, scene transition, and counterfactual inference.
MVBench reports a Score in ratio units across its video tasks.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- MVBench
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about MVBench.
Which model scores highest on MVBench?
GLM-5.3-Flash is currently ranked first with 77.80%.
What are the top three models on MVBench?
The current leaders are GLM-5.3-Flash (77.80%), Qwen3.5-122B-A10B (76.60%), and Qwen3.6-27B (75.50%).
Which MVBench model has the lowest official input price?
GLM-5.3-Flash has the lowest matched official input price at $0.08 input / $0.25 output per 1M tokens.
Which models are fastest among MVBench results?
The fastest matched records are GLM-5.3-Flash (74.28 tok/s via FriendliAI) and Qwen3 VL 4B Thinking (8.77 tok/s via DeepInfra).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.