multimodal benchmark
VideoMMMU evaluates Large Multimodal Models' ability to acquire knowledge from expert-level professional videos across six disciplines through perception, comprehension, and adaptation.
Updated Sep 5, 2026
Higher score ranks better on this benchmark.
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelGO | Score87.60% | Percentile100.00% | Participants26 | EvidenceC | Evaluated |
| Rank02 | ModelGO | Score86.90% | Percentile96.00% | Participants26 | EvidenceC | Evaluated |
| Rank03 | ModelMA | Score86.60% | Percentile92.00% | Participants26 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score85.90% | Percentile88.00% | Participants26 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score85.40% | Percentile84.00% | Participants26 | EvidenceC | Evaluated |
| Rank06 | ModelGO | Score84.80% | Percentile80.00% | Participants26 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score84.60% | Percentile76.00% | Participants26 | EvidenceC | Evaluated |
| Rank08 | ModelMI | Score84.60% | Percentile72.00% | Participants26 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score84.40% | Percentile68.00% | Participants26 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score84.00% | Percentile64.00% | Participants26 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score83.70% | Percentile60.00% | Participants26 | EvidenceC | Evaluated |
| Rank12 | ModelGO | Score83.60% | Percentile56.00% | Participants26 | EvidenceC | Evaluated |
| Rank13 | ModelOP | Score83.30% | Percentile52.00% | Participants26 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score82.30% | Percentile48.00% | Participants26 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score82.00% | Percentile44.00% | Participants26 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score80.40% | Percentile40.00% | Participants26 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score80.00% | Percentile36.00% | Participants26 | EvidenceC | Evaluated |
| Rank18 | ModelAC | Score79.00% | Percentile32.00% | Participants26 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score75.00% | Percentile28.00% | Participants26 | EvidenceC | Evaluated |
| Rank20 | ModelAC | Score74.70% | Percentile24.00% | Participants26 | EvidenceC | Evaluated |
| Rank21 | ModelAC | Score72.80% | Percentile20.00% | Participants26 | EvidenceC | Evaluated |
| Rank22 | ModelAC | Score69.40% | Percentile16.00% | Participants26 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score68.70% | Percentile12.00% | Participants26 | EvidenceC | Evaluated |
| Rank24 | ModelAC | Score65.30% | Percentile8.00% | Participants26 | EvidenceC | Evaluated |
| Rank25 | ModelOP | Score61.20% | Percentile4.00% | Participants26 | EvidenceC | Evaluated |
| Rank26 | ModelAC | Score56.20% | Percentile0.00% | Participants26 | EvidenceC | Evaluated |
The leading models and scores on this benchmark.
A closer view of the leading scores on this benchmark.
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What VideoMMMU measures and how its scores work.
VideoMMMU is a multimodal benchmark containing 300 videos and 900 human-annotated questions spanning Art, Business, Science, Medicine, Humanities, and Engineering.
It measures Large Multimodal Models' ability to acquire knowledge from expert-level professional videos across the cognitive stages of perception, comprehension, and adaptation, reported as a Score in ratio units.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
Common questions about VideoMMMU.
Gemini 3 Pro is currently ranked first with 87.60%.
The current leaders are Gemini 3 Pro (87.60%), Gemini 3 Flash (86.90%), and Kimi K2.5 (86.60%).
Qwen3.6-35B-A3B has the lowest matched official input price at $0.25 input / $1.5 output per 1M tokens.
The fastest matched records are GPT-4o (132.00 tok/s via OpenAI), Gemini 3 Flash (124.32 tok/s via Google), and GPT-5.2 (100.00 tok/s via OpenAI).
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.
It measures Large Multimodal Models' ability to acquire knowledge from expert-level professional videos across the cognitive stages of perception, comprehension, and adaptation, reported as a Score in ratio units.
Yes. Higher values rank better for this benchmark.
26 model results are currently shown.
No. This benchmark is shown for reference but does not contribute to the overall score.
Ranking basisThis videommmu AI model leaderboard uses descending score in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
Selection summary
Gemini 3 Pro currently leads VideoMMMU with 87.60%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price, and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.