multimodal benchmark
MME-RealWorld
A comprehensive evaluation benchmark for Multimodal Large Language Models featuring over 13,366 high-resolution images and 29,429 question-answer pairs across 43 subtasks and 5 real-world scenarios. The largest manually annotated multimodal benchmark to date, designed to test MLLMs on challenging high-resolution real-world scenarios.
Updated Aug 7, 2026
Published models1
Registry coverage1
MetricScore
EvidenceB
MME-RealWorld leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AC | 61.6% | 100.0% | 1 | C |
MME-RealWorld highlights
The top published results on this benchmark's own scale.
What is MME-RealWorld?
Definition and scoring fields from the benchmark registry.
A comprehensive evaluation benchmark for Multimodal Large Language Models featuring over 13,366 high-resolution images and 29,429 question-answer pairs across 43 subtasks and 5 real-world scenarios. The largest manually annotated multimodal benchmark to date, designed to test MLLMs on challenging high-resolution real-world scenarios.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MME-RealWorld
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- mme-realworld|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MME-RealWorld.
Which model scores highest on MME-RealWorld?
Qwen2.5-Omni-7B is currently ranked first with 61.6%.
What does MME-RealWorld measure?
A comprehensive evaluation benchmark for Multimodal Large Language Models featuring over 13,366 high-resolution images and 29,429 question-answer pairs across 43 subtasks and 5 real-world scenarios. The largest manually annotated multimodal benchmark to date, designed to test MLLMs on challenging high-resolution real-world scenarios.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.