Best AI Models for Math

Q: Which AI model is best at math?

Math rankings are updated live. Check the leaderboard above for the current leader on MATH-500, GSM8K, and competition-level math benchmarks.

Q: What is MATH-500?

MATH-500 is a curated subset of 500 competition-level math problems spanning multiple difficulty levels and topics including algebra, number theory, and geometry.

Q: Can AI solve competition math problems?

Top models now score above 95% on MATH-500 and are approaching human-competitive levels on AIME and Olympiad-style problems.

AI models ranked by math benchmarks. Compare MATH-500, GSM8K, and competition-level math scores across all providers.

148

Models

Providers

Open Source

$0.66

Median $/1M in

Top 3

Full Rankings

#	Model	Avg	frontierma	GSM8K	MATH level 5	otis mock	$/1M in	Context
1	Claude Instant🇺🇸 Anthropic	78.0	-	86.7	-	-	N/A	0K
2	Claude Fable 5🇺🇸 Anthropic	74.4	-	-	-	99.7	$10.00	1.0M
3	Qwen3.7 Max🇨🇳 Alibaba QwenOpen	68.1	-	-	-	95.0	$1.25	1.0M
4	GPT-5.4 Pro🇺🇸 OpenAI	63.2	50.0	-	-	-	$30.00	1.1M
5	Kimi K2.7 Code🇨🇳 moonshotaiOpen	63.1	-	-	-	96.4	$0.61	262K
6	Gemini 3.5 Flash🇺🇸 Google DeepMind	62.6	39.0	-	-	95.5	$1.50	1.0M
7	Claude Opus 4.8🇺🇸 Anthropic	61.4	47.2	-	-	98.3	$5.00	1.0M
8	Qwen-14B🇨🇳 Alibaba QwenOpen	60.7	-	61.3	-	-	N/A	0K
9	Gemini 3 Pro🇺🇸 Google DeepMind	60.4	37.6	-	-	91.4	N/A	0K
10	Qwen3.6 Plus🇨🇳 Alibaba QwenOpen	59.9	26.2	-	-	90.5	$0.33	1.0M
11	GLM 5.1🇨🇳 z-aiOpen	59.8	33.5	-	-	92.2	$0.97	203K
12	DeepSeek V3🇨🇳 DeepSeekOpen	59.0	1.7	-	64.8	15.8	$0.20	131K
13	U internlm-20b Unknown	59.0	-	62.9	-	-	N/A	0K
14	GPT-5.4🇺🇸 OpenAI	58.9	47.6	-	-	95.3	$2.50	1.1M
15	phi-3-medium 14B🇺🇸 MicrosoftOpen	58.6	-	-	17.6	-	N/A	0K
16	Qwen3 Max🇨🇳 Alibaba QwenOpen	58.3	-	-	97.1	73.3	$0.78	262K
17	Claude Opus 4.7🇺🇸 Anthropic	58.2	43.8	-	-	97.8	$5.00	1.0M
18	Falcon 2 11B TIIOpen	58.0	-	53.8	-	-	N/A	0K
19	R1 0528🇨🇳 DeepSeekOpen	57.9	-	-	96.6	66.4	$0.50	164K
20	Mixtral 8x7B🇫🇷 Mistral AIOpen	57.8	-	74.4	9.9	-	N/A	0K
21	Mixtral 8x7B Instruct🇫🇷 Mistral AIOpen	57.8	-	74.4	9.9	-	$0.54	33K
22	Claude Opus 4.6🇺🇸 Anthropic	57.1	40.7	-	-	94.4	$5.00	1.0M
23	Gemini 3.1 Pro Preview🇺🇸 Google DeepMind	56.6	36.9	-	-	95.6	$2.00	1.0M
24	Qwen3 235B A22B🇨🇳 Alibaba QwenOpen	56.4	-	-	68.9	-	$0.46	131K
25	GPT-5.2 Pro🇺🇸 OpenAI	56.2	-	-	-	-	$21.00	400K
26	GLM 5🇨🇳 z-aiOpen	56.1	16.4	-	-	80.0	$0.60	203K
27	Qwen3 235B A22B Thinking 2507🇨🇳 Alibaba QwenOpen	55.9	8.5	-	-	86.7	$0.15	262K
28	U Muse Spark Unknown	55.5	39.0	-	-	88.9	N/A	0K
29	GPT-4 (older v0314)🇺🇸 OpenAI	55.0	-	92.0	-	0.5	$30.00	8K
30	Grok 4🇺🇸 xAI	54.8	19.7	-	-	84.0	N/A	0K
31	GPT-5 Mini🇺🇸 OpenAI	54.3	27.2	-	97.8	86.7	$0.25	400K
32	o3🇺🇸 OpenAI	53.9	18.7	-	97.8	83.9	$2.00	200K
33	Gemini 2.0 Pro🇺🇸 Google DeepMind	53.7	-	-	83.5	-	N/A	0K
34	GPT-5.2🇺🇸 OpenAI	53.6	40.7	-	-	96.1	$1.75	400K
35	U Nemotron-4 15B Unknown	53.4	-	46.0	-	-	N/A	0K
36	Grok-3 mini🇺🇸 xAI	53.3	5.9	-	90.9	77.8	N/A	0K
37	Kimi K2 Thinking🇨🇳 moonshotaiOpen	53.3	21.4	-	-	83.0	$0.60	262K
38	GPT-5🇺🇸 OpenAI	53.2	32.4	-	98.1	91.4	$1.25	400K
39	Qwen2.5 Coder 32B Instruct🇨🇳 Alibaba QwenOpen	53.1	-	91.1	-	-	$0.66	128K
40	Gemini 2.5 Pro🇺🇸 Google DeepMind	52.7	14.1	-	95.6	84.7	$1.25	1.0M
41	Qwen2.5-Coder-3B🇨🇳 Alibaba QwenOpen	52.2	-	75.7	-	-	N/A	0K
42	Kimi K2.5🇨🇳 moonshotaiOpen	52.0	27.9	-	-	92.2	$0.38	262K
43	Kimi K2.6🇨🇳 moonshotaiOpen	51.7	39.0	-	-	96.1	$0.66	262K
44	Qwen2.5 72B Instruct🇨🇳 Alibaba QwenOpen	51.6	-	-	63.2	8.0	$0.36	131K
45	DeepSeek V3.2🇨🇳 DeepSeekOpen	51.2	22.1	-	-	87.8	$0.21	131K
46	Qwen3.6 Max Preview🇨🇳 Alibaba QwenOpen	51.2	23.1	-	-	91.1	$1.04	262K
47	GPT-4o (2024-05-13)🇺🇸 OpenAI	51.1	-	-	51.0	6.2	$5.00	128K
48	GPT-4 Turbo🇺🇸 OpenAI	51.0	-	90.0	23.0	1.0	$10.00	128K
49	o4 Mini🇺🇸 OpenAI	50.9	24.8	-	97.8	81.7	$1.10	200K
50	DeepSeek-Coder-V2-Lite-Base🇨🇳 DeepSeekOpen	50.8	-	67.1	-	-	N/A	0K
51	GLM 4.6🇨🇳 z-aiOpen	50.8	3.8	-	-	-	$0.43	203K
52	GLM 4.7🇨🇳 z-aiOpen	50.5	2.4	-	-	83.3	$0.40	203K
53	Qwen2.5 Coder 14B Instruct🇨🇳 AlibabaOpen	50.0	-	88.7	-	-	N/A	0K
54	Qwen3.6 Flash🇨🇳 Alibaba QwenOpen	49.8	10.3	-	-	86.1	$0.19	1.0M
55	GPT-5.1🇺🇸 OpenAI	49.6	31.0	-	-	88.6	$1.25	400K
56	o1🇺🇸 OpenAI	49.6	9.3	-	94.7	73.3	$15.00	200K
57	Claude Sonnet 4.6🇺🇸 Anthropic	48.0	32.4	-	-	85.8	$3.00	1.0M
58	Gemini 2.0 Flash🇺🇸 Google DeepMind	48.0	1.7	-	82.2	31.0	$0.10	1.0M
59	Gemini 2.0 Flash (Dec 2024)🇺🇸 Google DeepMind	48.0	1.7	-	82.2	31.0	N/A	0K
60	U Stable Beluga 2 Unknown	47.8	-	69.6	-	-	N/A	0K
61	Claude 3.7 Sonnet🇺🇸 Anthropic	47.7	4.1	-	91.2	57.7	N/A	0K
62	Gemini 3 Flash Preview🇺🇸 Google DeepMind	47.7	35.6	-	-	92.8	$0.50	1.0M
63	Gemini 1.5 Flash (May 2024)🇺🇸 Google DeepMind	47.3	0.0	82.4	25.1	3.8	N/A	0K
64	gpt-oss-120b🇺🇸 OpenAIOpen	46.9	-	-	-	88.9	$0.04	131K
65	GPT-3.5 Turbo (older v0613)🇺🇸 OpenAI	45.8	-	57.8	11.6	-	$1.00	4K
66	Mistral Large 2411🇫🇷 Mistral AIOpen	45.8	0.3	-	50.3	7.7	$2.00	131K
67	Grok 3🇺🇸 xAI	45.5	3.8	-	88.8	55.5	N/A	0K
68	R1🇨🇳 DeepSeekOpen	45.1	-	-	93.0	53.3	$0.70	164K
69	GPT-4.1 Mini🇺🇸 OpenAI	44.5	4.5	-	87.3	44.7	$0.40	1.0M
70	Falcon-180B TIIOpen	44.4	-	54.4	-	-	N/A	0K
71	Qwen2.5 Coder 7B Instruct🇨🇳 AlibabaOpen	44.4	-	86.7	-	-	N/A	0K
72	Qwen2.5 Coder 7B Instruct🇨🇳 Alibaba QwenOpen	44.4	-	86.7	-	-	$0.03	33K
73	Claude Opus 4.5🇺🇸 Anthropic	43.6	20.7	-	-	86.1	$5.00	200K
74	Claude Sonnet 4🇺🇸 Anthropic	43.3	4.1	-	84.4	71.1	$3.00	1.0M
75	GPT-4.1🇺🇸 OpenAI	43.3	5.5	-	83.0	38.3	$2.00	1.0M
76	GPT-5 Nano🇺🇸 OpenAI	43.3	8.3	-	95.2	81.1	$0.05	400K
77	GPT-4o-mini (2024-07-18)🇺🇸 OpenAI	43.2	-	91.3	52.6	6.8	$0.15	128K
78	Phi 4🇺🇸 MicrosoftOpen	43.2	-	-	64.9	13.7	$0.07	16K
79	Llama 2-13B🇺🇸 MetaOpen	42.5	-	36.9	3.3	-	N/A	0K
80	Claude 3.5 Sonnet🇺🇸 Anthropic	42.3	1.0	-	51.7	6.4	N/A	0K
81	Gemma 3 27B🇺🇸 Google DeepMindOpen	42.2	-	-	74.0	19.6	$0.08	131K
82	Gemma 3 27B (free)🇺🇸 Google DeepMindOpen	42.2	-	-	74.0	19.6	Free	131K
83	GPT-5 Pro🇺🇸 OpenAI	42.2	-	-	-	-	$15.00	400K
84	Claude Opus 4🇺🇸 Anthropic	41.7	4.5	-	85.0	64.4	$15.00	200K
85	Mistral 7B V0.1🇫🇷 Mistral AIOpen	41.6	-	54.4	-	-	N/A	0K
86	Gemini 1.5 Pro (May 2024)🇺🇸 Google DeepMind	41.5	-	-	40.8	6.7	N/A	0K
87	o1-preview🇺🇸 OpenAI	41.5	-	-	81.7	31.0	N/A	0K
88	Gemini 1.5 Pro (Feb 2024)🇺🇸 Google DeepMind	41.3	-	-	40.8	6.7	N/A	0K
89	Qwen2.5-Max🇨🇳 Alibaba QwenOpen	41.0	1.0	-	67.2	16.0	N/A	0K
90	U Baichuan 2-7B Unknown	40.3	-	24.6	-	-	N/A	0K
91	Gemma 2 27B🇺🇸 Google DeepMindOpen	40.1	-	84.9	27.9	1.3	$0.65	8K
92	Mistral Medium 3🇫🇷 Mistral AIOpen	40.0	0.3	-	81.6	32.1	$0.40	131K
93	Claude Sonnet 4.5🇺🇸 Anthropic	39.6	15.2	-	97.7	77.8	$3.00	1.0M
94	GPT-4o-mini🇺🇸 OpenAI	39.6	-	91.3	52.6	6.8	$0.15	128K
95	Mistral Large 2407🇫🇷 Mistral AIOpen	39.1	-	-	44.8	8.4	$2.00	131K
96	Qwen2-72B🇨🇳 Alibaba QwenOpen	39.0	-	-	39.1	-	N/A	0K
97	Qwen2.5 Coder 1.5B Instruct🇨🇳 AlibabaOpen	38.8	-	65.8	-	-	N/A	0K
98	Gemini 2.5 Flash🇺🇸 Google DeepMind	38.5	4.8	-	-	73.0	$0.30	1.0M
99	GPT-5.4 Mini🇺🇸 OpenAI	38.3	28.3	-	-	87.2	$0.75	400K
100	Llama 3.1 405B🇺🇸 MetaOpen	38.0	-	-	49.8	9.6	N/A	0K
101	Llama 3.1 70B Instruct🇺🇸 MetaOpen	37.8	-	-	36.7	3.5	$0.40	131K
102	Gemini 2.0 Flash Thinking (Jan 2025)🇺🇸 Google DeepMind	37.7	-	-	-	57.7	N/A	0K
103	Claude 2🇺🇸 Anthropic	37.2	-	-	11.7	2.4	N/A	0K
104	Claude 3.5 Haiku🇺🇸 Anthropic	37.2	0.3	-	46.4	4.2	N/A	0K
105	Mistral Nemo🇫🇷 Mistral AIOpen	37.2	-	84.2	10.8	-	$0.02	131K
106	Claude Opus 4.1🇺🇸 Anthropic	37.0	7.2	-	-	68.9	$15.00	200K
107	GPT-4o (2024-11-20)🇺🇸 OpenAI	36.4	0.3	-	53.3	6.3	$2.50	128K
108	Llama 3.2 90B🇺🇸 MetaOpen	36.1	-	-	39.4	2.5	N/A	0K
109	Gemma 2 9B🇺🇸 Google DeepMindOpen	36.0	-	84.9	21.0	0.5	$0.03	8K
110	GPT-4.5🇺🇸 OpenAI	35.9	-	-	78.6	37.7	N/A	0K
111	GPT-4o (2024-08-06)🇺🇸 OpenAI	35.6	0.3	-	53.3	6.3	$2.50	128K
112	GPT-4.1 Nano🇺🇸 OpenAI	35.2	1.0	-	70.0	28.8	$0.10	1.0M
113	o3 Mini🇺🇸 OpenAI	35.2	12.4	-	96.5	76.9	$1.10	200K
114	Qwen3.5-Flash🇨🇳 Alibaba QwenOpen	35.0	6.2	-	-	85.5	$0.07	1.0M
115	LLaMA-13B🇺🇸 MetaOpen	34.9	-	20.6	-	-	N/A	0K
116	o1-mini🇺🇸 OpenAI	34.9	1.7	-	89.2	46.9	N/A	0K
117	Qwen 3.5 Plus (hosted 397B-A17B)🇨🇳 Alibaba QwenOpen	34.8	21.0	-	-	85.0	N/A	0K
118	Claude Haiku 4.5🇺🇸 Anthropic	34.3	5.9	-	96.4	66.6	$1.00	200K
119	Claude 3 Opus🇺🇸 Anthropic	33.7	-	-	37.5	4.6	N/A	0K
120	GPT-5.4 Nano🇺🇸 OpenAI	33.6	25.9	-	-	87.8	$0.20	400K
121	Grok-2 (Dec 2024)🇺🇸 xAI	33.2	0.7	-	63.5	11.4	N/A	0K
122	U vicuna-13b-v1.1 Unknown	32.5	-	28.1	-	-	N/A	0K
123	Llama 3 70B Instruct🇺🇸 MetaOpen	32.4	-	-	22.6	4.2	$0.51	8K
124	U MPT-30B Unknown	31.7	-	34.4	-	-	N/A	0K
125	U Yi 6B UnknownOpen	31.4	-	44.9	5.2	-	N/A	0K
126	Llama 3 8B Instruct🇺🇸 MetaOpen	30.8	-	-	6.1	0.7	$0.14	8K
127	Mistral Large🇫🇷 Mistral AIOpen	30.0	0.3	-	24.5	1.9	$2.00	128K
128	Gemma 2B🇺🇸 Google DeepMindOpen	29.1	-	17.7	-	-	N/A	0K
129	Llama 3.3 70B Instruct (free)🇺🇸 MetaOpen	29.1	-	-	41.6	5.0	Free	131K
130	Claude 3 Haiku🇺🇸 Anthropic	28.7	-	-	14.9	1.7	$0.25	200K
131	Qwen-1_8B🇨🇳 Alibaba QwenOpen	28.7	-	21.2	-	-	N/A	0K
132	Claude 3 Sonnet🇺🇸 Anthropic	28.3	-	-	18.2	2.4	N/A	0K
133	Llama 4 Maverick🇺🇸 MetaOpen	28.0	0.7	-	73.0	20.5	$0.15	1.0M
134	Llama 3.1 8B Instruct🇺🇸 MetaOpen	27.4	-	82.4	22.9	2.4	$0.02	131K
135	DeepSeek Coder 33B🇨🇳 DeepSeekOpen	25.4	-	35.4	-	-	N/A	0K
136	Llama 3-70B🇺🇸 MetaOpen	24.7	-	-	22.6	4.2	N/A	0K
137	U StarCoder 2 15B UnknownOpen	24.3	-	57.7	-	-	N/A	0K
138	U Baichuan1-7B Unknown	23.7	-	9.2	-	-	N/A	0K
139	Mixtral 8x22B Instruct🇫🇷 Mistral AIOpen	23.5	-	-	24.2	-	$2.00	66K
140	CodeQwen1.5-7B🇨🇳 Alibaba QwenOpen	23.1	-	37.7	-	-	N/A	0K
141	Gemini 1.0 Pro🇺🇸 Google DeepMind	21.1	-	-	11.2	1.0	N/A	0K
142	Claude 2.1🇺🇸 Anthropic	21.0	-	-	-	1.9	N/A	0K
143	Qwen2.5 Coder 0.5B Instruct🇨🇳 AlibabaOpen	20.8	-	34.5	-	-	N/A	0K
144	U INTELLECT-1 Unknown	20.2	-	38.6	-	-	N/A	0K
145	Llama 4 Scout🇺🇸 MetaOpen	18.9	0.0	-	62.3	7.7	$0.10	10.0M
146	DeepSeek Coder 6.7B🇨🇳 DeepSeekOpen	16.7	-	21.3	-	-	N/A	0K
147	U Magistral Small 1.1 Unknown	16.5	-	-	-	29.9	N/A	0K
148	DeepSeek Coder 1.3B🇨🇳 DeepSeekOpen	3.2	-	4.4	-	-	N/A	0K

90+ Gold 80-89 70-79 60-69 <60Scores in % unless noted. Avg = unweighted mean across tested benchmarks.

About this category

Models ranked by mathematical ability across MATH-500, GSM8K, FrontierMath, and competition-level problems. Scores reflect arithmetic, algebra, geometry, combinatorics, and proof-based reasoning.

Related categories

Best AI Models for Reasoning

AI models ranked by reasoning benchmarks. Compare GPQA Diamond, ARC-AGI, BBH, and other reasoning tests across all providers.

Best AI Models for Coding

AI models ranked by coding benchmarks. Compare HumanEval+, SWE-bench Verified, Aider Polyglot, and more across all providers.

Best AI Models for Knowledge

AI models ranked by knowledge benchmarks. Compare MMLU-Pro, GPQA Diamond, SimpleQA, and other knowledge tests.

Frequently asked questions

Which AI model is best at math?

Math rankings are updated live. Check the leaderboard above for the current leader on MATH-500, GSM8K, and competition-level math benchmarks.

What is MATH-500?

MATH-500 is a curated subset of 500 competition-level math problems spanning multiple difficulty levels and topics including algebra, number theory, and geometry.

Can AI solve competition math problems?

Top models now score above 95% on MATH-500 and are approaching human-competitive levels on AIME and Olympiad-style problems.

Back to all models