MMLU-Redux
MMLU-Redux - cleaned and verified subset of MMLU.
Category
Knowledge
Max Score
100.0
Score Type
percent
Active
Yes
Model Scores
Scores for MMLU-Redux
| Model | Score | Date | Verified |
|---|---|---|---|
| Qwen3.6-35B-A3B |
90.34%
|
16.04.2026 | Verified |
| Gemma4-31B-it |
92.02%
|
16.04.2026 | Unverified |
| Qwen3.5-35B-A3B |
90.34%
|
16.04.2026 | Unverified |
| Gemma4-26B-A4B |
87.82%
|
16.04.2026 | Unverified |
| Qwen3.6-27B |
91.18%
|
22.04.2026 | Verified |
| Qwen3.5-397B-A17B |
97.06%
|
22.04.2026 | Unverified |
| Claude 4.5 Opus |
100.00%
|
22.04.2026 | Unverified |
| Qwen3.5-27B |
89.92%
|
16.04.2026 | Unverified |
| Qwen3.5-4B |
71.43%
|
01.02.2026 | Verified |
| Qwen3.5-122B-A10B |
93.28%
|
01.02.2026 | Verified |
| Qwen3-235B-A22B |
92.44%
|
01.02.2026 | Unverified |
| GPT-5-mini 2025-08-07 |
92.02%
|
01.02.2026 | Unverified |
| Gemini-2.5-Flash-Thinking |
85.29%
|
31.07.2025 | Verified |
| Qwen3-30B-A3B |
74.37%
|
29.04.2025 | Verified |
| Qwen3-30B-A3B-Thinking-2507 |
82.35%
|
31.07.2025 | Verified |
| DeepSeek-V3.2 |
65.97%
|
26.06.2026 | Verified |
| DeepSeek-V4-Flash |
73.95%
|
26.06.2026 | Verified |
| DeepSeek-V4-Pro |
79.83%
|
26.06.2026 | Verified |
| LFM2.5-2.6B |
34.45%
|
— | Verified |
| Qwen3.5-2B |
34.45%
|
— | Verified |
| Gemma4-E2B |
71.80
|
— | Verified |
| granite-4.2-3B |
29.83%
|
— | Verified |
| Gemma4-E4B |
50.00%
|
— | Verified |
| LFM2.5-8B-A1B |
34.45%
|
— | Verified |
| Qwen3-Next-80B-A3B-Thinking |
86.97%
|
10.09.2025 | Verified |
| Qwen3-VL-30B-A3B |
69.75%
|
04.10.2025 | Verified |
| MiniCPM5-2B |
54.20%
|
— | Verified |
| Nemotron-3-Nano-4B |
33.61%
|
— | Verified |