AIME

Artificial Analysis · 50 models

Rank	Model	Score
1	GPT-5 (High)	0.9567
2	Grok 4	0.9433
3	Qwen3 235B A22B 2507 (Reasoning)	0.94
3	o4-mini (High)	0.94
5	Grok 3 mini Reasoning (High)	0.9333
6	GPT-5 (Medium)	0.9167
7	Qwen3 30B A3B 2507 (Reasoning)	0.9067
8	o3	0.9033
9	DeepSeek R1 0528 (May '25)	0.8933
10	Gemini 2.5 Pro	0.8867
11	GLM-4.5 (Reasoning)	0.8733
12	Gemini 2.5 Pro Preview (Mar' 25)	0.87
13	Llama Nemotron Super 49B v1.5 (Reasoning)	0.86
13	o3-mini (High)	0.86
15	MiniMax M1 80k	0.8467
16	EXAONE 4.0 32B (Reasoning)	0.8433
16	Gemini 2.5 Flash Preview (Reasoning)	0.8433
16	Gemini 2.5 Pro Preview (May' 25)	0.8433
19	Qwen3 235B A22B (Reasoning)	0.84
20	GPT-5 (Low)	0.83
21	Gemini 2.5 Flash (Reasoning)	0.8233
22	MiniMax M1 40k	0.8133
23	Qwen3 32B (Reasoning)	0.8067
24	Sonar Reasoning Pro	0.79
25	QwQ 32B	0.78
26	Claude 4 Sonnet (Reasoning)	0.7733
27	Sonar Reasoning	0.77
27	o3-mini	0.77
29	Qwen3 14B (Reasoning)	0.7633
30	Claude 4 Opus (Reasoning)	0.7567
31	Qwen3 30B A3B (Reasoning)	0.7533
32	Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)	0.7467
32	Qwen3 8B (Reasoning)	0.7467
34	Qwen3 30B A3B 2507 Instruct	0.7267
35	o1	0.7233
36	Qwen3 235B A22B 2507 Instruct	0.7167
37	Magistral Small 1	0.7133
38	Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning)	0.7067
39	Gemini 2.5 Flash-Lite (Reasoning)	0.7033
40	Magistral Medium 1	0.7
41	Kimi K2	0.6933
42	Solar Pro 2 (Reasoning)	0.69
43	DeepSeek R1 Distill Qwen 32B	0.6867
44	DeepSeek R1 (Jan '25)	0.6833
45	GLM-4.5-Air	0.6733
46	DeepSeek R1 Distill Llama 70B	0.67
47	DeepSeek R1 Distill Qwen 14B	0.6667
48	Solar Pro 2 (Preview) (Reasoning)	0.6633
49	Qwen3 4B (Reasoning)	0.6567
50	DeepSeek R1 0528 Qwen3 8B	0.65