MMLU-CF 0-shot Test: leaderboard

MMLU-CF 0-shot Test — Official MMLU-CF contamination-free benchmark; 0-shot test cohort. Original MMLU column and test-minus-validation deltas are not independent capability panels. Native API/checkpoint labels retained. Metric: Accuracy (%). Source: https://github.com/microsoft/MMLU-CF

Metric: Accuracy (%). Source: github.com. Saturation forecast: Estimated already saturated. 44 models tracked. Updated 16 Nov 2026.

Top models

#ModelScoreOverall rank
1GPT-4o71.9#744
2Qwen 2.5 72B Instruct70.6#1095
3Llama 3.1 70B Instruct70.4#1542
4GPT-4 Turbo68.9#999
5Qwen 2.5 32B Instruct68.9#1232
6phi-4-14B68.5#1666
7Llama 3 70B Instruct68.1#1689
8Llama 3.3 70B Instruct67.6#1379
9Qwen 2.5 14B Instruct67#1750
10GPT-4o Mini66#1658
11Mixtral 8x22B Instruct65.3#2097
12Gemma 2 27B64.2#2141
13Phi-3.5-MoE-instruct63.1#2199
14Phi-3-medium-128k-instruct62.5#3321
15Qwen 2 72B Instruct62.4#1514

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=mmlu-cf-0-shot-test · How It Works · Data refreshed daily, snapshot 2026-11-16.