multimodal benchmark
MotionBench
MotionBench is a benchmark for evaluating multimodal models on motion understanding in videos, testing the ability to comprehend temporal dynamics, movement patterns, and action sequences.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
MotionBench leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | BY | 74.9% | 100.0% | 3 | C | |
| 2 | BY | 74.8% | 50.0% | 3 | C | |
| 3 | MA | 70.4% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
MotionBench highlights
The top published results on this benchmark's own scale.
What is MotionBench?
Definition and scoring fields from the benchmark registry.
MotionBench is a benchmark for evaluating multimodal models on motion understanding in videos, testing the ability to comprehend temporal dynamics, movement patterns, and action sequences.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MotionBench
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- motionbench|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MotionBench.
Which model scores highest on MotionBench?
Seed 2.1 Pro is currently ranked first with 74.9%.
What does MotionBench measure?
MotionBench is a benchmark for evaluating multimodal models on motion understanding in videos, testing the ability to comprehend temporal dynamics, movement patterns, and action sequences.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 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.