Capability ranking

Best AI for Math

Best Math Models

Updated 2026-08-11

Ranked models109

Vendors19

Open-weight models23

Official prices46

GPT-5.2-Pro leads this page at 99.1, 0.0 points ahead of Grok 4 Heavy.

MiMo V2.5 Pro is the highest-ranked open-weight option at #5. Nova Micro has the lowest official input price among ranked models.

Top math modelGPT-5.2-Pro99.1 math score Best open-weight modelMiMo V2.5 ProRank #5 Lowest official inputNova Micro$0.04 / 1M Largest contextGemini 1.5 Pro2.1M tokens

Top math models

ModelReasoningKnowledgeMathCodingInstruction

GPT-5.2-ProOpenAI78.559.899.1N/AN/A

Grok 4 HeavyxAI88.280.499.186.1N/A

Gemini 2.0 FlashGoogle33.760.097.227.861.9

Qwen3.7 MaxAlibaba Cloud / Qwen Team91.589.193.185.692.2

MiMo V2.5 ProXiaomi84.062.892.849.7N/A

Math scoreOverall score

80 models with comparable dataOpen a model by selecting its logo

GPQA93.2%GPT-5.2 Pro96 compared models

MMLU-Pro89.6%Qwen3.7 Max69 compared models

AIME 2025100.0%GPT-5.251 compared models

SWE-Bench Verified80.6%DeepSeek-V4-Pro-Max42 compared models

MMLU91.8%o150 compared models

Humanity's Last Exam54.7%GLM-5.239 compared models

LiveCodeBench93.5%DeepSeek-V4-Pro-Max34 compared models

MATH97.9%o3-mini38 compared models


01	OPGPT-5.2-ProOpenAI	GPT-5.2 Pro	99.1	71.6	78.5	59.8	N/A	N/A	60.0%	400K	Closed weights	$21	$168
02	XAGrok 4 HeavyxAI	Grok-4 Heavy	99.1	68.2	88.2	80.4	86.1	N/A	80.0%	N/A	Closed weights	N/A	N/A
03	GOGemini 2.0 FlashGoogle	Gemini 2.0 Flash	97.2	28.1	33.7	60.0	27.8	61.9	100.0%	1M	Closed weights	N/A	N/A
04	ACQwen3.7 MaxAlibaba Cloud / Qwen Team	Qwen3.7 Max	93.1	78.8	91.5	89.1	85.6	92.2	100.0%	1M	Closed weights	$2.5	$7.5
05	XIMiMo V2.5 ProXiaomi	MiMo-V2.5-Pro	92.8	46.7	84.0	62.8	49.7	N/A	80.0%	1M	Open weights	$0.435	$0.87
06	ZAGLM 4.5Zhipu AI	GLM-4.5	90.0	49.3	61.4	51.3	49.6	N/A	80.0%	131.1K	Open weights	$0.60	$2.2
07	OPGPT-5.2OpenAI	GPT-5.2	90.0	71.4	74.0	69.9	87.5	N/A	80.0%	400K	Closed weights	$1.75	$14
08	ZAGLM 5.2Zhipu AI	GLM-5.2	89.8	79.3	92.7	88.6	56.0	N/A	80.0%	1M	Open weights	$1.4	$4.4
09	GOGemma 3 27BGoogle	Gemma 3 27B	89.0	15.8	46.8	37.9	39.2	61.1	100.0%	131.1K	Closed weights	N/A	N/A
10	DEDeepSeek-V3.2DeepSeek	DeepSeek-V3.2-Speciale	88.5	58.1	N/A	54.4	54.1	N/A	60.0%	131.1K	Closed weights	$0.14	$0.28
11	ZAGLM 4.5 AirZhipu AI	GLM-4.5-Air	86.5	41.5	54.3	39.6	30.9	N/A	80.0%	131.1K	Open weights	$0.20	$1.1
12	GOGemini 2.0 Flash LiteGoogle	Gemini 2.0 Flash-Lite	86.0	21.0	26.6	51.4	50.0	0.0	100.0%	1M	Closed weights	N/A	N/A
13	DEDeepSeek-V4-ProDeepSeek	DeepSeek-V4-Pro-Max	83.9	73.8	90.1	84.8	63.7	N/A	80.0%	1M	Closed weights	$0.435	$0.87
14	ACQwen2.5 32BAlibaba Cloud / Qwen Team	Qwen2.5 32B Instruct	81.7	14.8	50.6	37.4	68.3	N/A	80.0%	N/A	Closed weights	$0.70	$2.8
15	ACQwen2.5 72BAlibaba Cloud / Qwen Team	Qwen2.5 72B Instruct	80.1	21.4	22.5	37.5	60.3	68.4	100.0%	131.1K	Closed weights	$1.4	$5.6
16	ANClaude Sonnet 3.5Anthropic	Claude 3.5 Sonnet	79.9	32.1	77.8	52.0	55.7	N/A	80.0%	200K	Closed weights	N/A	N/A
17	MELlama 3.1 405BMeta	Llama 3.1 405B Instruct	75.0	24.5	67.2	41.4	75.4	62.5	100.0%	128K	Closed weights	N/A	N/A
18	GOGemma 3 12BGoogle	Gemma 3 12B	74.8	11.2	39.4	35.4	32.4	51.4	100.0%	131.1K	Closed weights	N/A	N/A
19	SASarvam 105BSarvam AI	Sarvam-105B	73.4	44.6	60.1	42.0	31.7	40.6	100.0%	131.1K	Open weights	N/A	N/A
20	ACQwen3.6 PlusAlibaba Cloud / Qwen Team	Qwen3.6 Plus	73.3	67.5	80.9	87.3	51.8	72.7	100.0%	1M	Closed weights	$0.50	$3
21	ACQwen2.5 14BAlibaba Cloud / Qwen Team	Qwen2.5 14B Instruct	73.1	8.2	20.3	40.0	51.2	N/A	80.0%	N/A	Closed weights	$0.35	$1.4
22	MAKimi K2.6Moonshot AI	Kimi K2.6	72.7	75.5	92.3	58.7	71.2	N/A	80.0%	262.1K	Open weights	$0.95	$4
23	XAGrok 4xAI	Grok-4	72.3	56.5	51.9	64.1	83.3	N/A	80.0%	256K	Closed weights	N/A	N/A
24	STStep 3.5 FlashStepFun	Step-3.5-Flash	72.0	60.4	N/A	N/A	60.4	N/A	40.0%	256K	Open weights	$0.10	$0.30
25	AMNova ProAmazon	Nova Pro	69.9	20.9	68.6	65.2	75.4	86.7	100.0%	300K	Closed weights	$0.80	$3.2
26	DEDeepSeek-V2.5DeepSeek	DeepSeek-V2.5	69.4	15.4	63.6	40.4	38.2	81.8	100.0%	8.2K	Closed weights	N/A	N/A
27	NVNemotron 3 SuperNVIDIA	Nemotron 3 Super (120B A12B)	68.7	44.8	70.0	60.8	31.9	71.4	100.0%	262.1K	Open weights	$0.20	$0.80
28	ZAGLM 5.1Zhipu AI	GLM-5.1	68.1	69.6	80.7	84.8	53.8	N/A	80.0%	200K	Open weights	$1.4	$4.4
29	MAKimi K2.5Moonshot AI	Kimi K2.5	67.3	66.0	84.1	87.3	47.3	N/A	80.0%	262.1K	Open weights	$0.60	$3
30	ACQwen3.7 PlusAlibaba Cloud / Qwen Team	Qwen3.7-Plus	66.9	73.8	84.7	83.5	61.6	87.8	100.0%	1M	Closed weights	$0.50	$3
31	GOGemini 1.5 ProGoogle	Gemini 1.5 Pro	66.9	20.5	62.0	46.1	41.5	75.7	100.0%	2.1M	Closed weights	N/A	N/A
32	MAKimi K2 ThinkingMoonshot AI	Kimi K2-Thinking-0905	65.7	61.0	75.5	82.0	61.0	0.0	100.0%	262.1K	Closed weights	$0.60	$2.5
33	XAGrok 4 FastxAI	Grok 4 Fast	65.2	54.0	79.4	67.4	88.9	N/A	80.0%	2M	Closed weights	N/A	N/A
34	OPGPT-5.1OpenAI	GPT-5.1	63.9	62.1	86.5	N/A	71.2	N/A	60.0%	400K	Closed weights	$1.25	$10
35	OPGPT-5.1-InstantOpenAI	GPT-5.1 Instant	63.9	61.3	86.5	N/A	71.2	N/A	60.0%	400K	Closed weights	N/A	N/A
36	OPo3 miniOpenAI	o3-mini	63.8	36.3	58.1	46.9	51.7	70.0	100.0%	200K	Closed weights	$1.1	$4.4
37	AMNova LiteAmazon	Nova Lite	63.7	13.2	51.4	41.9	48.5	71.9	100.0%	300K	Closed weights	$0.06	$0.24
38	ACQwen3 235B A22B ThinkingAlibaba Cloud / Qwen Team	Qwen3-235B-A22B-Thinking-2507	63.5	48.5	71.4	61.9	53.9	84.9	100.0%	262.1K	Closed weights	N/A	N/A
39	ZAGLM 4.7Zhipu AI	GLM-4.7	60.6	59.7	79.4	74.5	47.2	N/A	80.0%	204.8K	Open weights	$0.60	$2.2
40	SASarvam 30BSarvam AI	Sarvam-30B	59.1	34.3	39.1	57.8	59.8	N/A	80.0%	128K	Open weights	N/A	N/A
41	MIMAI Thinking 1Microsoft	MAI-Thinking-1	58.8	55.1	74.5	71.1	45.8	38.4	100.0%	N/A	Closed weights	N/A	N/A
42	ACQwen2.5 7BAlibaba Cloud / Qwen Team	Qwen2.5 7B Instruct	58.4	6.5	9.4	17.2	37.6	35.0	100.0%	131.1K	Closed weights	$0.175	$0.70
43	OPo1OpenAI	o1	57.2	36.3	59.4	81.7	36.0	N/A	80.0%	200K	Closed weights	$15	$60
44	XIMiMo V2 FlashXiaomi	MiMo-V2-Flash	57.0	51.3	73.0	63.0	46.5	N/A	80.0%	262.1K	Open weights	$0.14	$0.28
45	NVNemotron Nano 9BNVIDIA	Nemotron Nano 9B v2	56.6	32.4	34.3	N/A	75.0	79.7	80.0%	131.1K	Open weights	N/A	N/A
46	ANClaude Opus 3Anthropic	Claude 3 Opus	55.4	16.7	73.8	30.9	46.1	N/A	80.0%	200K	Closed weights	N/A	N/A
47	ACQwen3 VL 235B A22B ThinkingAlibaba Cloud / Qwen Team	Qwen3 VL 235B A22B Thinking	53.8	45.8	58.6	55.9	46.1	77.7	100.0%	262.1K	Closed weights	N/A	N/A
48	GOGemini 1.5 FlashGoogle	Gemini 1.5 Flash	52.4	10.7	55.1	28.1	23.1	37.7	100.0%	1M	Closed weights	N/A	N/A
49	GOGemma 3 4BGoogle	Gemma 3 4B	51.8	0.0	21.7	13.3	10.8	50.4	100.0%	131.1K	Closed weights	N/A	N/A
50	NVLlama 3.1 Nemotron Ultra 253BNVIDIA	Llama 3.1 Nemotron Ultra 253B v1	51.6	30.5	57.1	N/A	66.7	68.8	80.0%	N/A	Closed weights	N/A	N/A
51	MIMiniMax M1 80KMiniMax	MiniMax M1 80K	51.0	37.0	45.1	32.7	42.5	37.5	100.0%	1M	Closed weights	N/A	N/A
52	MELongCat Flash ThinkingMeituan	LongCat-Flash-Thinking-2601	50.9	60.8	63.5	48.9	68.4	N/A	80.0%	128K	Closed weights	N/A	N/A
53	AMNova MicroAmazon	Nova Micro	50.7	4.3	40.1	29.3	34.6	47.7	100.0%	128K	Closed weights	$0.035	$0.14
54	ACQwen3.5 397B A17BAlibaba Cloud / Qwen Team	Qwen3.5-397B-A17B	50.1	66.3	81.3	82.8	57.4	81.9	100.0%	262.1K	Open weights	$0.60	$3.6
55	ACQwen3.6 27BAlibaba Cloud / Qwen Team	Qwen3.6-27B	49.9	60.5	71.8	71.4	49.0	N/A	80.0%	262.1K	Open weights	$0.60	$3.6
56	OPGPT-4.5OpenAI	GPT-4.5	48.7	36.2	43.8	91.0	33.4	38.7	100.0%	128K	Closed weights	N/A	N/A
57	MAKimi k1.5Moonshot AI	Kimi-k1.5	47.9	30.1	N/A	61.4	N/A	47.7	60.0%	N/A	Closed weights	N/A	N/A
58	ACQwen3 Next 80B A3B ThinkingAlibaba Cloud / Qwen Team	Qwen3-Next-80B-A3B-Thinking	45.7	40.9	58.2	65.2	38.5	62.7	100.0%	131.1K	Open weights	$0.50	$6
59	OPGPT-5-miniOpenAI	GPT-5 mini	45.0	44.9	68.5	28.3	N/A	N/A	60.0%	400K	Closed weights	$0.25	$2
60	ACQwen2.5 Omni 7BAlibaba Cloud / Qwen Team	Qwen2.5-Omni-7B	44.3	10.2	5.2	7.8	26.1	100.0	100.0%	N/A	Closed weights	$0.10	$0.40
61	DEDeepSeek-R1-Distill-QwenDeepSeek	DeepSeek R1 Distill Qwen 7B	43.7	15.7	23.2	N/A	29.9	N/A	60.0%	N/A	Closed weights	N/A	N/A
62	MELongCat Flash LiteMeituan	LongCat-Flash-Lite	43.5	33.0	40.3	54.7	21.0	N/A	80.0%	256K	Closed weights	N/A	N/A
63	ACQwen2 72BAlibaba Cloud / Qwen Team	Qwen2 72B Instruct	41.4	9.8	50.9	24.9	46.6	N/A	80.0%	N/A	Closed weights	N/A	N/A
64	ACQwen3.6 35B A3BAlibaba Cloud / Qwen Team	Qwen3.6-35B-A3B	40.9	55.3	80.3	61.6	32.3	N/A	80.0%	262.1K	Open weights	$0.248	$1.49
65	MIMiniMax M1 40KMiniMax	MiniMax M1 40K	40.9	35.3	43.4	30.1	38.6	37.5	100.0%	N/A	Closed weights	N/A	N/A
66	NVLlama 3.3 Nemotron Super 49BNVIDIA	Llama-3.3 Nemotron Super 49B v1	39.3	23.0	39.5	N/A	96.9	90.9	80.0%	131.1K	Open weights	N/A	N/A
67	ACQwen2.5 Coder 32BAlibaba Cloud / Qwen Team	Qwen2.5-Coder 32B Instruct	39.3	4.4	46.3	17.2	68.0	N/A	80.0%	128K	Closed weights	$0.287	$0.861
68	ACQwen3 VL 30B A3B ThinkingAlibaba Cloud / Qwen Team	Qwen3 VL 30B A3B Thinking	38.8	31.8	37.8	43.4	30.8	39.9	100.0%	262.1K	Closed weights	N/A	N/A
69	DEDeepSeek-R1DeepSeek	DeepSeek-R1-0528	38.6	44.2	65.2	70.3	30.0	N/A	80.0%	131.1K	Closed weights	N/A	N/A
70	MELongCat Flash ChatMeituan	LongCat-Flash-Chat	38.2	36.9	49.3	75.0	49.6	70.3	100.0%	128K	Closed weights	N/A	N/A
71	OPo3OpenAI	o3	37.3	52.6	47.3	22.3	64.5	67.9	100.0%	200K	Closed weights	$2	$8
72	ANClaude Sonnet 3Anthropic	Claude 3 Sonnet	35.7	3.5	58.4	18.0	20.0	N/A	80.0%	200K	Closed weights	N/A	N/A
73	MIPhi 4 MiniMicrosoft	Phi 4 Mini	35.6	0.0	23.8	48.2	N/A	N/A	60.0%	128K	Open weights	$0.075	$0.30
74	ANClaude Sonnet 3.7Anthropic	Claude 3.7 Sonnet	34.9	38.7	76.8	50.0	45.5	90.6	100.0%	200K	Closed weights	N/A	N/A
75	XAGrok 1.5xAI	Grok-1.5	34.5	0.0	9.0	13.3	21.5	N/A	80.0%	N/A	Closed weights	N/A	N/A
76	MAKimi K2Moonshot AI	Kimi K2-Instruct-0905	34.5	37.2	55.1	38.0	35.3	63.4	100.0%	N/A	Closed weights	$0.60	$2.5
77	MIPhi 3.5 MoEMicrosoft	Phi-3.5-MoE-instruct	34.3	4.9	61.8	50.6	39.7	75.0	100.0%	N/A	Closed weights	N/A	N/A
78	ACQwen3 VL 8B ThinkingAlibaba Cloud / Qwen Team	Qwen3 VL 8B Thinking	33.6	28.5	32.5	47.0	28.9	41.3	100.0%	262.1K	Closed weights	N/A	N/A
79	ACQwQ 32BAlibaba Cloud / Qwen Team	QwQ-32B	31.6	24.8	36.5	N/A	59.7	34.4	80.0%	N/A	Closed weights	$0.287	$0.861
80	OPGPT-5-nanoOpenAI	GPT-5 nano	29.5	31.1	47.2	9.8	N/A	N/A	60.0%	400K	Closed weights	$0.05	$0.40
81	ACQwen3 VL 235B A22BAlibaba Cloud / Qwen Team	Qwen3 VL 235B A22B Instruct	28.0	43.2	54.0	64.8	53.2	76.8	100.0%	262.1K	Closed weights	N/A	N/A
82	DEDeepSeek-R1-Distill-LlamaDeepSeek	DeepSeek R1 Distill Llama 8B	27.9	12.4	22.5	N/A	31.9	N/A	60.0%	N/A	Closed weights	N/A	N/A
83	GOGemini 1.5 Flash 8BGoogle	Gemini 1.5 Flash 8B	27.5	0.0	11.2	18.8	28.6	8.3	100.0%	1M	Closed weights	N/A	N/A
84	NVLlama 3.1 Nemotron Nano 8BNVIDIA	Llama 3.1 Nemotron Nano 8B V1	27.0	10.4	28.8	N/A	84.4	17.7	80.0%	N/A	Closed weights	N/A	N/A
85	MIMAI Code 1 FlashMicrosoft	MAI-Code-1-Flash	26.7	50.4	76.2	33.7	32.2	64.3	100.0%	256K	Closed weights	N/A	N/A
86	IBGranite 3.3 8BIBM	Granite 3.3 8B Instruct	25.3	3.8	27.4	48.7	80.8	10.2	100.0%	128K	Closed weights	N/A	N/A
87	DEDeepSeek-V3DeepSeek	DeepSeek-V3 0324	24.8	26.0	42.1	66.4	36.1	N/A	80.0%	163.8K	Closed weights	N/A	N/A
88	MIPhi 3.5 miniMicrosoft	Phi-3.5-mini-instruct	24.4	0.0	27.7	42.9	18.7	100.0	100.0%	128K	Closed weights	N/A	N/A
89	ANClaude Haiku 3Anthropic	Claude 3 Haiku	23.1	0.0	46.4	24.8	27.7	N/A	80.0%	200K	Closed weights	N/A	N/A
90	ACQwen2 7BAlibaba Cloud / Qwen Team	Qwen2 7B Instruct	22.4	0.0	3.0	11.9	17.1	45.5	100.0%	N/A	Closed weights	N/A	N/A
91	ACQwen3 235B A22BAlibaba Cloud / Qwen Team	Qwen3-235B-A22B-Instruct-2507	21.5	42.4	61.1	70.6	51.2	72.8	100.0%	262.1K	Open weights	$0.70	$2.8
92	ACQwen3 VL 4B ThinkingAlibaba Cloud / Qwen Team	Qwen3 VL 4B Thinking	20.9	23.0	28.4	30.2	13.5	29.2	100.0%	262.1K	Closed weights	N/A	N/A
93	ACQwen2.5 Coder 7BAlibaba Cloud / Qwen Team	Qwen2.5-Coder 7B Instruct	19.9	0.0	22.8	3.7	50.5	N/A	80.0%	N/A	Closed weights	$0.144	$0.287
94	GOGemma 3n E4BGoogle	Gemma 3n E4B Instructed	19.9	0.0	8.0	23.4	16.9	29.6	100.0%	32K	Closed weights	N/A	N/A
95	GOGemma 3n E4B LiteRTGoogle	Gemma 3n E4B Instructed LiteRT Preview	19.9	0.0	25.7	26.4	16.9	29.6	100.0%	N/A	Closed weights	N/A	N/A
96	ACQwen3 Next 80B A3BAlibaba Cloud / Qwen Team	Qwen3-Next-80B-A3B-Instruct	18.9	36.6	53.4	53.0	49.0	62.7	100.0%	131.1K	Open weights	$0.50	$2
97	MELlama 3.2 3BMeta	Llama 3.2 3B Instruct	16.7	0.0	23.2	4.0	N/A	14.1	80.0%	128K	Closed weights	N/A	N/A
98	ACQwen3 VL 30B A3BAlibaba Cloud / Qwen Team	Qwen3 VL 30B A3B Instruct	14.3	28.8	30.6	41.6	7.7	37.6	100.0%	262.1K	Closed weights	N/A	N/A
99	GOGemma 2 27BGoogle	Gemma 2 27B	10.7	0.0	68.6	59.4	8.6	N/A	80.0%	N/A	Closed weights	N/A	N/A
100	DEDeepSeek-V3.1DeepSeek	DeepSeek-V3.1	7.5	39.9	53.6	66.3	36.8	N/A	80.0%	131.1K	Open weights	N/A	N/A
101	GOGemma 3 1BGoogle	Gemma 3 1B	6.9	0.0	0.1	0.6	1.0	11.7	100.0%	N/A	Closed weights	N/A	N/A
102	GOGemma 3n E2BGoogle	Gemma 3n E2B Instructed	6.5	0.0	1.8	11.0	7.0	11.4	100.0%	N/A	Closed weights	N/A	N/A
103	GOGemma 3n E2B LiteRTGoogle	Gemma 3n E2B Instructed LiteRT (Preview)	6.5	0.0	9.5	10.3	7.0	11.4	100.0%	N/A	Closed weights	N/A	N/A
104	OPGPT-4.1-miniOpenAI	GPT-4.1 mini	5.3	24.8	35.6	38.6	13.1	21.1	100.0%	1M	Closed weights	$0.40	$1.6
105	GOGemma 2 9BGoogle	Gemma 2 9B	5.3	0.0	49.9	39.5	2.3	N/A	80.0%	N/A	Closed weights	N/A	N/A
106	ACQwen3 VL 8BAlibaba Cloud / Qwen Team	Qwen3 VL 8B Instruct	5.2	24.8	22.4	25.4	3.9	36.0	100.0%	262.1K	Closed weights	N/A	N/A
107	ACQwen3 VL 4BAlibaba Cloud / Qwen Team	Qwen3 VL 4B Instruct	4.0	19.3	12.3	33.4	1.9	16.8	100.0%	262.1K	Closed weights	N/A	N/A
108	OPGPT-4.1OpenAI	GPT-4.1	3.5	31.9	38.6	49.0	32.3	29.3	100.0%	1M	Closed weights	$2	$8
109	BAERNIE 4.5Baidu	ERNIE 4.5	0.0	0.0	6.5	0.3	0.0	0.0	100.0%	128K	Closed weights	N/A	N/A

Official PAYG prices appear here. Third-party offers remain on the pricing and model detail pages.

Math scoreOfficial API price blend (8:1 input/output), USD / 1M

Efficient frontier46 models with official PAYG prices

Top-model vendor mix

25 models

Alibaba Cloud / Qwen Team5

Access model

Open weights

Closed weights

Official price available

GPT-5.2-ProvsGrok 4 Heavy GPT-5.2-ProvsMiMo V2.5 Pro GPT-5.2-ProvsNova Micro

Common questions about Best AI for Math.

What is the best model on Best AI for Math?

GPT-5.2-Pro is currently ranked first with a math score of 99.1.

How is this leaderboard ranked?

Each model appears once using its current scored version. Category pages use the capability named in the title, while the overall page uses the LLMBoard score.

How many models are included?

This page currently ranks 109 unique model products.

Where does pricing come from?

Main price columns use the model vendor's official standard PAYG API rate. Eligible third-party offers appear only in separately labeled columns, and unavailable official prices display as N/A.

Does Arena affect the score?

No. Arena results are displayed as an independent signal and are not included in the current LLMBoard capability score.