MMLU-Pro

Artificial Analysis · 50 models

Rank	Model	Score
1	Gemini 3 Pro Preview (High)	0.898
2	Claude Opus 4.5 (Reasoning)	0.895
2	Gemini 3 Pro Preview (Low)	0.895
4	Gemini 3 Flash Preview (Reasoning)	0.89
5	Claude Opus 4.5 (Non-reasoning)	0.889
6	Gemini 3 Flash Preview (Non-reasoning)	0.882
7	Claude 4.1 Opus (Reasoning)	0.88
8	Claude 4.5 Sonnet (Reasoning)	0.875
8	MiniMax-M2.1	0.875
10	GPT-5.2 (Xhigh)	0.874
11	Claude 4 Opus (Reasoning)	0.873
12	GPT-5 (High)	0.871
13	GPT-5.1 (High)	0.87
14	GPT-5 (Medium)	0.867
15	Grok 4	0.866
16	GPT-5 Codex (High)	0.865
17	DeepSeek V3.2 Speciale	0.863
18	DeepSeek V3.2 (Reasoning)	0.862
18	Gemini 2.5 Pro	0.862
20	Claude 4 Opus (Non-reasoning)	0.86
20	Claude 4.5 Sonnet (Non-reasoning)	0.86
20	GPT-5 (Low)	0.86
20	GPT-5.1 Codex (High)	0.86
24	GPT-5.2 (Medium)	0.859
25	Gemini 2.5 Pro Preview (Mar' 25)	0.858
26	GLM-4.7 (Reasoning)	0.856
27	Doubao Seed Code	0.854
27	Grok 4.1 Fast (Reasoning)	0.854
29	o3	0.853
30	DeepSeek V3.1 (Reasoning)	0.851
30	DeepSeek V3.1 Terminus (Reasoning)	0.851
32	DeepSeek V3.2 Exp (Reasoning)	0.85
32	Grok 4 Fast (Reasoning)	0.85
34	Cogito v2.1 (Reasoning)	0.849
34	DeepSeek R1 0528 (May '25)	0.849
36	Kimi K2 Thinking	0.848
36	o1-preview	0.848
38	DeepSeek R1 (Jan '25)	0.844
39	MiMo-V2-Flash (Reasoning)	0.843
39	Qwen3 235B A22B 2507 (Reasoning)	0.843
41	Claude 4 Sonnet (Reasoning)	0.842
41	Gemini 2.5 Flash Preview (Sep '25) (Reasoning)	0.842
43	Qwen3 Max	0.841
43	o1	0.841
45	K-EXAONE (Reasoning)	0.838
45	Qwen3 Max (Preview)	0.838
47	Claude 3.7 Sonnet (Reasoning)	0.837
47	Claude 4 Sonnet (Non-reasoning)	0.837
47	DeepSeek V3.2 (Non-reasoning)	0.837
47	GPT-5 mini (High)	0.837