Math

AIME-2025

American Invitational Mathematics Examination 2025.

33 models published a score

#	Model	Company	Score
1	GPT-5.2	OpenAI	100.0
2	Grok 4 Heavy	xAI	100.0
3	Qwen3-Max-Thinking	Alibaba	100.0
4	Gemini 3 Flash	Google DeepMind	99.7
5	Step 3.5 Flash	StepFun	97.3
6	MAI-Thinking-1	Microsoft	97.0
7	Kimi K2.5	Moonshot AI	96.1
8	DeepSeek V3.2 Speciale	DeepSeek	96.0
9	GLM-4.7	Zhipu AI	95.7
10	Claude Sonnet 4.6	Anthropic	95.6
11	Gemini 3 Pro	Google DeepMind	95.0
12	Doubao Seed 2.0 Pro	ByteDance	93.3
13	DeepSeek V3.2	DeepSeek	93.1
14	Doubao Seed 2.0 Lite	ByteDance	93.0
15	EXAONE 4.5 33B	LG AI Research	92.9
16	K-EXAONE 236B-A23B	LG AI Research	92.8
17	Qwen3.6-35B-A3B	Alibaba	92.7
18	Qwen3.5-397B-A17B	Alibaba	91.3
19	Nova 2 Lite	Amazon	91.0
20	Nemotron 3 Super	Nvidia	90.2
21	Gemma 4 (31B dense)	Google DeepMind	89.2
22	Nemotron 3 Nano	Nvidia	89.1
23	Gemma 4 26B-A4B	Google DeepMind	88.3
24	DeepSeek R1 0528	DeepSeek	87.5
25	Mistral Medium 3.5	Mistral AI	86.3
26	GPT-5.5 Instant	OpenAI	81.2
27	Qwen3-Max	Alibaba	80.6
28	Step-3	StepFun	73.0
29	Magistral Medium 1.2	Mistral AI	65.0
30	Gemma 4 E4B	Google DeepMind	42.5
31	Mistral Large 3	Mistral AI	40.0
32	Gemma 4 E2B	Google DeepMind	37.5
33	Reka Flash 3	Reka	33.7