MLVU
MLVU - multi-task long video understanding benchmark.
Category
Video Understanding
Max Score
100.0
Score Type
percent
Active
Yes
Model Scores
Scores for MLVU
| Model | Score | Date | Verified |
|---|---|---|---|
| Qwen3.6-35B-A3B |
92.41%
|
16.04.2026 | Verified |
| Qwen3.5-35B-A3B |
88.28%
|
16.04.2026 | Unverified |
| Claude-Sonnet-4.5 |
72.80
|
16.04.2026 | Unverified |
| Qwen3.6-27B |
95.17%
|
22.04.2026 | Verified |
| Qwen3.5-397B-A17B |
95.86%
|
22.04.2026 | Unverified |
| Claude 4.5 Opus |
61.38%
|
22.04.2026 | Unverified |
| Qwen3.5-27B |
90.34%
|
16.04.2026 | Unverified |
| Qwen3.5-4B |
68.97%
|
01.02.2026 | Verified |
| Qwen3.5-122B-A10B |
100.00%
|
01.02.2026 | Verified |
| GPT-5-mini 2025-08-07 |
72.41%
|
01.02.2026 | Unverified |
| Qwen3-235B-A22B |
75.86%
|
01.02.2026 | Unverified |
| Qwen3-VL-235B-A22B |
75.86%
|
01.02.2026 | Verified |