Benchmark
MMLU-Pro Benchmark
What is MMLU-Pro?
Static closed-set multiple choice; contamination risk rises over time. Benchmark Radar tracks 11 reported scores for MMLU-Pro, cited by 13 source documents…
- Released
- 2024-06-03
- Evaluates
- knowledge
- Openness
- unknown
- Importer
- Model reports
- Reported scores
- 11
Source provenance
- Original evidence https://github.com/TIGER-AI-Lab/MMLU-Pro
MMLU-Pro code
MMLU-Pro results and reported scores
Model reports
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| DeepSeek-R1 | DeepSeek | 84.0 | 2025-01-22 | evidence |
| DeepSeek-V3 | DeepSeek | 75.9 | 2024-12-27 | evidence |
| DeepSeek-V4-Flash | DeepSeek | 87.5 | 2026-06-25 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 82.9 | 2026-04-22 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 87.1 | 2026-04-22 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 87.5 | 2026-04-22 | evidence |
| Gemma 4 (31B) | 85.2 | 2026-03-11 | evidence | |
| Qwen3.5-397B-A17B | Qwen | 87.8 | 2026-02-16 | evidence |
| Seed2.0 Lite | ByteDance | 70.3 | 2026-02-14 | evidence |
| Seed2.0 Mini | ByteDance | 63.4 | 2026-02-14 | evidence |
| Seed2.0 Pro | ByteDance | 87.0 | 2026-02-14 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.
Which sources cite MMLU-Pro?
- DeepSeek-V4 DeepSeek, 2026-06-25
- DeepSeek-V4-Pro DeepSeek, 2026-04-22
- Gemma 4 (31B) Google, 2026-03-11
- Qwen3.5-397B-A17B Qwen, 2026-02-16
- Seed2.0 (Pro / Lite / Mini) ByteDance, 2026-02-14
- Olmo 3 (7B and 32B) Ai2, 2025-11-20
- Qwen3 Qwen, 2025-05-14
- Mistral Medium 3 Mistral, 2025-05-07
- Llama 4 Meta, 2025-04-05
- Claude 3.7 Sonnet Anthropic, 2025-02-24
- DeepSeek-R1 DeepSeek, 2025-01-22
- DeepSeek-V3 DeepSeek, 2024-12-27
13 source documents cite this benchmark.