Benchmark
MMLU-ProX 5-shot CoT BN
Scores are percentages. Higher is better.
- Models measured
- 7
- Scores recorded
- 8
- Sources
- 1
- IQ weight
- 0.21
Score distribution
- 69.7%
- Mean
- 69.8%
- Median
- 3.6%
- Std dev
- 66.6%
- 10th percentile
- 72.2%
- 90th percentile
- 66.4%
- Min
This benchmark has a weight of 0.21 for chat models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Qwen3 235B (Web)think· MMLU-ProX Performance Results docs | 77.8% |
| 2 | GPT 4.1· MMLU-ProX Performance Results docs | 72.2% |
| 3 | o4 Mini· MMLU-ProX Performance Results docs | 70.1% |
| 4 | DeepSeek V3· MMLU-ProX Performance Results docs | 69.8% |
| 5 | Qwen3 235B (Web)· MMLU-ProX Performance Results docs | 67.7% |
| 6 | Qwen3 14Bthink· MMLU-ProX Performance Results docs | 66.8% |
| 7 | DeepSeek R1· MMLU-ProX Performance Results docs | 66.6% |
| 8 | Qwen3 32Bthink· MMLU-ProX Performance Results docs | 66.4% |
Data credits: MMLU-ProX: Xuan et al., arXiv:2503.10497. Public project-page results; no results licence stated.