AIME 2025

AIME 2025: competition math; accuracy.

180models

–Open Index weight

–last updated

Artificial Analysispublisher

#	Model	Effort	Score	Cost / task	Harness
1	GPT-5.2	xhigh	99.0%	–	–
2	GPT-5 Codex	high	98.7%	–	–
3	Gemini 3 Flash	–	97.0%	–	–
4	DeepSeek V3.2	–	96.7%	–	–
5	MiMo V2 Flash	–	96.3%	–	–
6	Gemini 3 Pro	high	95.7%	–	–
7	GPT-5.1 Codex	high	95.7%	–	–
8	GLM 4.7	–	95.0%	–	–
9	Kimi K2	–	94.7%	–	–
10	KAT-Coder-Pro V1	none	94.7%	–	–
11	GPT-5	high	94.3%	–	–
12	Nova 2.0 Lite (High)	high	94.3%	–	–
13	GPT-5.1	high	94.0%	–	–
14	gpt-oss-120b	high	93.4%	–	–
15	Grok 4	–	92.7%	–	–
16	GPT-5.1 Codex Mini	high	91.7%	–	–
17	Opus 4.5	–	91.3%	–	–
18	Qwen 3 235B-A22B	–	91.0%	–	–
19	NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)	–	91.0%	–	–
20	GPT-5 Mini	high	90.7%	–	–
21	o4 mini	high	90.7%	–	–
22	K-EXAONE (Reasoning)	–	90.3%	–	–
23	Grok 4 Fast	–	89.7%	–	–
24	DeepSeek V3.1	–	89.7%	–	–
25	Nova 2.0 Omni (Medium)	medium	89.7%	–	–
26	Grok 4.1 Fast	–	89.3%	–	–
27	Ring-1T	–	89.3%	–	–
28	gpt-oss-20b	high	89.3%	–	–
29	Nova 2.0 Pro Preview (Medium)	medium	89.0%	–	–
30	Nova 2.0 Lite (Medium)	medium	88.7%	–	–
31	o3	–	88.3%	–	–
32	Qwen 3	–	88.3%	–	–
33	Sonnet 4.5	–	88.0%	–	–
34	Apriel-v1.6-15B-Thinker	–	88.0%	–	–
35	INTELLECT-3 (Based on GLM-4.5-Air)	–	88.0%	–	–
36	Gemini 2.5 Pro	–	87.7%	–	–
37	Apriel-v1.5-15B-Thinker	–	87.5%	–	–
38	GLM 4.6	–	86.0%	–	–
39	GLM 4.6 V	–	85.3%	–	–
40	ERNIE 5.0 Thinking Preview	–	85.0%	–	–
41	Grok 3 Mini	high	84.7%	–	–
42	Seed-OSS-36B-Instruct	–	84.7%	–	–
43	Haiku 4.5	–	83.7%	–	–
44	GPT-5 Nano	high	83.7%	–	–
45	Ring-flash-2.0	–	83.7%	–	–
46	MiniMax M2.1	–	82.7%	–	–
47	Qwen 3 4B	–	82.7%	–	–
48	Qwen 3 Max	max	82.3%	–	–
49	Magistral Medium 1.2	medium	82.0%	–	–
50	GLM 4.5 Air	–	80.7%	–	–
51	Opus 4.1	–	80.3%	–	–
52	Motif-2-12.7B-Reasoning	–	80.3%	–	–
53	Magistral Small 1.2	–	80.3%	–	–
54	EXAONE 4.0 32B (Reasoning)	–	80.0%	–	–
55	Falcon-H1R-7B	–	80.0%	–	–
56	o1	–	79.7%	–	–
57	Doubao Seed Code	–	79.3%	–	–
58	Mi:dm K 2.5 Pro Preview	–	78.7%	–	–
59	MiniMax M2	–	78.3%	–	–
60	Gemini 2.5 Flash	–	78.3%	–	–
61	K2-V2 (High)	high	78.3%	–	–
62	Olmo 3.1 32B Think	–	77.3%	–	–
63	Mi:dm K 2.5 Pro	–	76.7%	–	–
64	Llama Nemotron Super 49B v1.5 (Reasoning)	–	76.7%	–	–
65	DeepSeek R1	–	76.0%	–	–
66	NVIDIA Nemotron Nano 12B v2 VL (Reasoning)	–	75.0%	–	–
67	Sonnet 4	–	74.3%	–	–
68	GLM 4.5	–	73.7%	–	–
69	Olmo 3 32B Think	–	73.7%	–	–
70	Opus 4	–	73.3%	–	–
71	GLM 4.5 V	–	73.0%	–	–
72	Qwen 3 32B	–	73.0%	–	–
73	Cogito v2.1 (Reasoning)	–	72.7%	–	–
74	Qwen 3 30B-A3B	–	72.3%	–	–
75	Ling-1T	none	71.3%	–	–
76	Olmo 3 7B Think	–	70.7%	–	–
77	Llama 3.1 405b	–	69.7%	–	–
78	NVIDIA Nemotron Nano 9B V2 (Reasoning)	–	69.7%	–	–
79	Gemini 2.5 Flash Lite	–	68.7%	–	–
80	Llama 3.1 70b	–	68.7%	–	–
81	Ling-flash-2.0	none	65.3%	–	–
82	K2-V2 (Medium)	medium	64.7%	–	–
83	Llama 3.1	max	63.7%	–	–
84	Nova 2.0 Pro Preview (Low)	low	63.3%	–	–
85	NVIDIA Nemotron Nano 9B V2 (Non-reasoning)	none	62.3%	–	–
86	Solar Pro 2 (Reasoning)	high	61.3%	–	–
87	MiniMax M1	–	61.0%	–	–
88	HyperCLOVA X SEED Think (32B)	–	59.0%	–	–
89	Grok 3	none	58.0%	–	–
90	Qwen 3 14B	none	58.0%	–	–
91	Sonnet 3.7	–	56.3%	–	–
92	Nova 2.0 Omni (Low)	low	56.0%	–	–
93	Llama 3.3	–	54.7%	–	–
94	Exaone 4.0 1.2B (Reasoning)	–	50.3%	–	–
95	Ling-mini-2.0	none	49.3%	–	–
96	Nova 2.0 Lite (Low)	low	46.7%	–	–
97	GPT-4.1 Mini	none	46.3%	–	–
98	K-EXAONE (Non-reasoning)	none	44.0%	–	–
99	Grok Code Fast 1	–	43.3%	–	–
100	Magistral Small 1	–	41.3%	–	–
101	ERNIE 4.5 300B A47B	none	41.3%	–	–
102	Olmo 3 7B Instruct	none	41.3%	–	–
103	DeepSeek V3	none	41.0%	–	–
104	Magistral Medium 1	medium	40.3%	–	–
105	Qwen 3 Coder	none	39.3%	–	–
106	EXAONE 4.0 32B (Non-reasoning)	none	39.3%	–	–
107	Qwen 3 1.7B	–	38.7%	–	–
108	Mistral Medium 3.1	none	38.3%	–	–
109	Mistral Large 3	none	38.0%	–	–
110	Nova 2.0 Omni (Non-reasoning)	none	37.0%	–	–
111	Devstral 2	none	36.7%	–	–
112	Kimi Linear 48B A3B Instruct	none	36.3%	–	–
113	K2-V2 (Low)	low	35.3%	–	–
114	GPT-4.1	none	34.7%	–	–
115	Devstral Small 2	none	34.3%	–	–
116	Nova 2.0 Lite (Non-reasoning)	none	33.7%	–	–
117	Reka Flash 3	–	33.7%	–	–
118	Ministral 3 8B	none	31.7%	–	–
119	Nova 2.0 Pro Preview (Non-reasoning)	none	30.7%	–	–
120	Mistral Medium 3	none	30.3%	–	–
121	Solar Pro 2 (Non-reasoning)	none	30.0%	–	–
122	Ministral 3 14B	none	30.0%	–	–
123	Devstral Small	none	29.3%	–	–
124	QwQ 32B	–	29.0%	–	–
125	Mistral Small 3.2	none	27.0%	–	–
126	NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)	none	26.7%	–	–
127	GPT-4o	none	25.7%	–	–
128	LFM2 8B A1B	none	25.3%	–	–
129	Qwen 3 8B	none	24.3%	–	–
130	GPT-4.1 Nano	none	24.0%	–	–
131	Exaone 4.0 1.2B (Non-reasoning)	none	24.0%	–	–
132	Ministral 3 3B	none	22.0%	–	–
133	Gemini 2.0 Flash	none	21.7%	–	–
134	Gemma 3 27B	none	20.7%	–	–
135	Llama 4 Maverick	none	19.3%	–	–
136	Gemma 3 12B	none	18.3%	–	–
137	Phi-4	none	18.0%	–	–
138	Qwen 3 0.6B	–	18.0%	–	–
139	Nova Premier	none	17.3%	–	–
140	GPT-4o mini	none	14.7%	–	–
141	Gemma 3n E4B Instruct	none	14.3%	–	–
142	Qwen 2.5 72B	none	14.0%	–	–
143	Mistral Large 2	none	14.0%	–	–
144	Llama 4 Scout	none	14.0%	–	–
145	Granite 4.0 H Small	none	13.7%	–	–
146	NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)	none	13.3%	–	–
147	Command A	none	13.0%	–	–
148	Gemma 3 4B	none	12.7%	–	–
149	Jamba Reasoning 3B	–	10.7%	–	–
150	Gemma 3n E2B Instruct	none	10.3%	–	–
151	LFM2 2.6B	none	8.3%	–	–
152	Llama Nemotron Super 49B v1.5 (Non-reasoning)	none	8.0%	–	–
153	Llama 3.3 70b	none	7.7%	–	–
154	Nova Pro	none	7.0%	–	–
155	Nova Lite	none	7.0%	–	–
156	Phi-4 Mini Instruct	none	6.7%	–	–
157	Granite 3.3 8B (Non-reasoning)	none	6.7%	–	–
158	Granite 4.0 H 1B	none	6.3%	–	–
159	Granite 4.0 1B	none	6.3%	–	–
160	Nova Micro	none	6.0%	–	–
161	Granite 4.0 Micro	none	6.0%	–	–
162	Devstral Medium	none	4.7%	–	–
163	Llama 3.1 8b	none	4.3%	–	–
164	Mistral Small 3	none	4.3%	–	–
165	Mistral Small 3.1	none	3.7%	–	–
166	OLMo 2 32B	none	3.3%	–	–
167	Llama 3.2 3b	none	3.3%	–	–
168	LFM2 1.2B	none	3.3%	–	–
169	Gemma 3 1B	none	3.3%	–	–
170	Pixtral Large	none	2.3%	–	–
171	Jamba 1.7 Large	none	2.3%	–	–
172	Gemma 3 270M	none	2.3%	–	–
173	Llama 3.2 11b	none	1.7%	–	–
174	Granite 4.0 H 350M	none	1.3%	–	–
175	OLMo 2 7B	none	0.7%	–	–
176	Phi-3 Mini Instruct 3.8B	none	0.3%	–	–
177	Jamba 1.7 Mini	none	0.3%	–	–
178	Molmo 7B-D	none	0.0%	–	–
179	Granite 4.0 350M	none	0.0%	–	–
180	Llama 3.2 1b	none	0.0%	–	–