multimodal benchmark
MMBench Leaderboard
A bilingual benchmark for assessing multi-modal capabilities of vision-language models through multiple-choice questions in both English and Chinese, providing systematic evaluation across diverse vision-language tasks with robust metrics.
Updated Aug 17, 2026
Models9
Model coverage9
MetricScore
EvidenceB
MMBench Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelST | Score91.8% | Percentile100.0% | Participants9 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score88.0% | Percentile87.5% | Participants9 | EvidenceC | Evaluated |
| Rank03 | ModelMI | Score86.7% | Percentile75.0% | Participants9 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score86.5% | Percentile62.5% | Participants9 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score84.3% | Percentile50.0% | Participants9 | EvidenceC | Evaluated |
| Rank06 | ModelMI | Score81.9% | Percentile37.5% | Participants9 | EvidenceC | Evaluated |
| Rank07 | ModelDE | Score80.3% | Percentile25.0% | Participants9 | EvidenceC | Evaluated |
| Rank08 | ModelDE | Score79.6% | Percentile12.5% | Participants9 | EvidenceC | Evaluated |
| Rank09 | ModelDE | Score69.2% | Percentile0.0% | Participants9 | EvidenceC | Evaluated |
MMBench Highlights
The leading models and scores on this benchmark.
MMBench Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for MMBench
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis mmbench AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01STStep3-VL-10BStepFun
- Score
- 91.8%
Strengths
- Ranks #1 of 9 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures MMBench, not total model capability
- 02ACQwen2.5 VL 72B InstructAlibaba Cloud / Qwen Team
- Score
- 88.0%
Strengths
- Ranks #2 of 9 compared models
- 88th percentile on this benchmark
- C evidence result
Considerations
- This result measures MMBench, not total model capability
- 03MIPhi-4-multimodal-instructMicrosoft
- Score
- 86.7%
- Speed
- Up to 25 tok/s via DeepInfra
Strengths
- Ranks #3 of 9 compared models
- 75th percentile on this benchmark
- C evidence result
Considerations
- This result measures MMBench, not total model capability
- 04ACQwen2-VL-72B-InstructAlibaba Cloud / Qwen Team
- Score
- 86.5%
Strengths
- Ranks #4 of 9 compared models
- 63th percentile on this benchmark
- C evidence result
Considerations
- This result measures MMBench, not total model capability
- 05ACQwen2.5 VL 7B InstructAlibaba Cloud / Qwen Team
- Score
- 84.3%
- Price
- $0.35 input / $1.1 output per 1M tokens
Strengths
- Ranks #5 of 9 compared models
- 50th percentile on this benchmark
- C evidence result
Considerations
- This result measures MMBench, not total model capability
Selection summary
Best AI Models for MMBench
Step3-VL-10B currently leads MMBench with 91.8%. 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.
What is MMBench?
What MMBench measures and how its scores work.
A bilingual benchmark for assessing multi-modal capabilities of vision-language models through multiple-choice questions in both English and Chinese, providing systematic evaluation across diverse vision-language tasks with robust metrics.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- MMBench
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- mmbench|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about MMBench.
Which model scores highest on MMBench?
Step3-VL-10B is currently ranked first with 91.8%.
What does MMBench measure?
A bilingual benchmark for assessing multi-modal capabilities of vision-language models through multiple-choice questions in both English and Chinese, providing systematic evaluation across diverse vision-language tasks with robust metrics.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
9 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.