How is this ranking computed?

It is the same family composite as the leaderboard. Each benchmark is weighted by how much of its spread survives its own measurement error.

Best AI for math

Ranked by the mathematics composite: MATH level 5, FrontierMath and OTIS Mock AIME. Full leaderboard.

Rank	Model	Lab	Composite	Evidence	Out $/M
1	Claude Fable 5	Anthropic	100th	14 bench · 2 src	$50.00
2	AI Co-Mathematician	Google	100th	2 bench · 1 src	—
3	GPT-6 Astra	OpenAI	99th	11 bench · 2 src	$50.00
4	GPT-6 Astra	OpenAI	99th	1 bench · 1 src	—
5	GPT-5.4 Pro	OpenAI	98th	1 bench · 1 src	—
6	GPT-5.6 Sol	OpenAI	98th	2 bench · 1 src	$10.00
7	Claude Fable 5.1	Anthropic	97th	10 bench · 2 src	$50.00
8	GPT-5.5 Pro	OpenAI	97th	2 bench · 1 src	$180.00
9	GPT-5.6 Sol	OpenAI	97th	12 bench · 2 src	$10.00
10	GPT-5.5 Pro pre-release	OpenAI	96th	5 bench · 1 src	—
11	Claude Opus 5	Anthropic	96th	9 bench · 2 src	$25.00
12	GPT-5.4	OpenAI	95th	13 bench · 2 src	$15.00
13	GPT-5.6 Terra	OpenAI	95th	8 bench · 2 src	$12.00
14	GPT-5.2 Pro	OpenAI	94th	1 bench · 1 src	—
15	Gemini 2.5 Pro Exp	Google	94th	2 bench · 1 src	—
16	GPT-5.5	OpenAI	94th	16 bench · 2 src	$30.00
17	GPT-5.4 Pro	OpenAI	93rd	6 bench · 1 src	$180.00
18	Qwen3.7 Plus	Alibaba	93rd	4 bench · 1 src	$3.072
19	Claude Opus 4.8	Anthropic	92nd	13 bench · 2 src	$25.00
20	gpt-oss-120b	OpenAI	92nd	6 bench · 3 src	$0.798
21	Qwen3.8 Max	Alibaba	92nd	8 bench · 2 src	$6.00
22	Nemotron 3 Ultra	NVIDIA	91st	4 bench · 1 src	$3.00
23	Gemini 3 Pro Preview	Google	91st	6 bench · 2 src	$9.60
24	Claude Opus 4.8	Anthropic	90th	3 bench · 1 src	—
25	GPT-5.6 Luna	OpenAI	90th	8 bench · 2 src	$0.37
26	Claude Opus 4.7	Anthropic	89th	13 bench · 2 src	$25.00
27	Gemma 4 26B-A4B IT	Google	89th	3 bench · 1 src	$0.29
28	Gemini 3.7 Flash	Google	89th	8 bench · 2 src	$3.75
29	Kimi K3	Kimi	88th	8 bench · 2 src	free
30	GPT-5.2 Pro	OpenAI	88th	2 bench · 1 src	$168.00
31	Claude Opus 4.6	Anthropic	87th	11 bench · 2 src	$25.00
32	Qwen3.8 Max (0902)	Alibaba	87th	6 bench · 1 src	—
33	Kimi K2.6	Kimi	86th	12 bench · 2 src	$4.00
34	Muse Spark	Meta	86th	4 bench · 1 src	—
35	Grok 4.6	xAI	86th	9 bench · 2 src	$6.00
36	Qwen3.7-Max	Alibaba	85th	10 bench · 2 src	$9.00
37	Gemini 3.1 Pro Preview	Google	85th	12 bench · 2 src	$12.00
38	Claude Sonnet 4.6	Anthropic	84th	8 bench · 2 src	$15.94
39	DeepSeek R1	DeepSeek	84th	4 bench · 2 src	$1.70
40	GLM-5.3	Zhipu	83rd	8 bench · 2 src	$4.40
41	DeepSeek V4 Pro 0813	DeepSeek	83rd	8 bench · 2 src	$4.0077
42	Gemini 3.8 Flash	Google	83rd	8 bench · 2 src	$3.75
43	GPT 5.2	OpenAI	82nd	12 bench · 3 src	$14.00
44	Gemma 4 31B IT	Google	82nd	4 bench · 1 src	$0.42
45	DeepSeek-R1-Distill-Llama-70B	DeepSeek	81st	3 bench · 1 src	$0.99
46	Claude Sonnet 5	Anthropic	81st	8 bench · 2 src	$10.00
47	Gemini 2.5 Flash	Google	80th	2 bench · 2 src	$2.50
48	Gemini 3.5 Flash	Google	80th	14 bench · 3 src	$9.00
49	Claude Opus 4.6	Anthropic	80th	4 bench · 1 src	—
50	MiniMax M3	MiniMax	79th	5 bench · 2 src	$2.40
51	DeepSeek-R1-Distill-Qwen-14B	DeepSeek	79th	4 bench · 1 src	$0.431
52	GLM-5.1	Zhipu	78th	8 bench · 1 src	$4.40
53	Grok 4.5	xAI	78th	7 bench · 2 src	$6.00
54	GPT-5 Pro	OpenAI	77th	4 bench · 1 src	$120.00
55	DeepSeek V4 Flash 0731	DeepSeek	77th	8 bench · 2 src	$0.50
56	GPT-5.6 Sol	OpenAI	77th	4 bench · 1 src	—
57	Qwen3.6 Max Preview	Alibaba	76th	7 bench · 1 src	free
58	GPT 5	OpenAI	76th	13 bench · 3 src	$10.00
59	DeepSeek-R1	DeepSeek	75th	4 bench · 1 src	—
60	seed-oss-36b-instruct	—	75th	3 bench · 1 src	$0.57
61	GLM-5.2	Zhipu	74th	10 bench · 2 src	$4.40
62	Qwen3.5 Plus	Alibaba	74th	7 bench · 1 src	$2.40
63	Qwen3-30B-A3B	Alibaba	74th	3 bench · 1 src	—
64	GPT 5.1	OpenAI	73rd	8 bench · 2 src	$10.00
65	Qwen3.6 27B	Alibaba	73rd	5 bench · 2 src	free
66	Kimi K2.5	Kimi	72nd	7 bench · 2 src	$1.90
67	GLM-5.3-Flash	Zhipu	72nd	7 bench · 2 src	$0.025
68	Qwen3 Max	Alibaba	71st	7 bench · 1 src	$6.001
69	Gemini 3.6 Flash	Google	71st	8 bench · 2 src	$3.75
70	Qwen3.6 Plus	Alibaba	71st	9 bench · 2 src	$3.00
71	Gemini 3 Flash Preview	Google	70th	10 bench · 2 src	$3.00
72	Kimi K2.7 Code	Kimi	70th	7 bench · 2 src	$4.389
73	Gemini 2.5 Pro Preview	Google	69th	4 bench · 1 src	—
74	Qwen3.5-35B-A3B	Alibaba	69th	3 bench · 1 src	$1.80
75	GPT-5.2	OpenAI	68th	4 bench · 1 src	—
76	Qwen3.5 9B	Alibaba	68th	3 bench · 1 src	—
77	Qwen3.5 397B-A17B	Alibaba	68th	4 bench · 1 src	$2.65
78	Qwen3.7 Plus	Alibaba	67th	5 bench · 1 src	—
79	Gemini 2.5 Deep Think	Google	67th	2 bench · 1 src	—
80	Grok 4.20	xAI	66th	6 bench · 1 src	$6.00
81	Inkling Small	Thinking Machines	66th	7 bench · 1 src	$1.20
82	GPT-4.5	OpenAI	65th	4 bench · 2 src	$150.00
83	Kimi K2 Instruct	Kimi	65th	7 bench · 3 src	free
84	QwQ 32B	Alibaba	65th	4 bench · 2 src	$0.861
85	Qwen3.5 397B-A17B	Alibaba	64th	5 bench · 1 src	—
86	glm-4.7-flash	—	64th	3 bench · 1 src	$0.40
87	Grok 4.3 Beta	xAI	63rd	6 bench · 1 src	—
88	GPT-5.5	OpenAI	63rd	4 bench · 1 src	—
89	GPT-5.4	OpenAI	62nd	4 bench · 1 src	—
90	GPT 5 mini	OpenAI	62nd	11 bench · 2 src	$2.00
91	Grok 4	xAI	62nd	6 bench · 2 src	$15.00
92	o3	OpenAI	61st	11 bench · 3 src	$8.00
93	Qwen3 8B	Alibaba	61st	3 bench · 1 src	$0.47
94	Qwen3.5-4B	Alibaba	60th	1 bench · 1 src	$0.20
95	DeepSeek-R1-Distill-Qwen-32B	DeepSeek	60th	3 bench · 1 src	$0.861
96	Grok-3 mini	xAI	59th	5 bench · 1 src	$0.50
97	DeepSeek v4	DeepSeek	59th	7 bench · 1 src	—
98	GPT-5.4 nano	OpenAI	59th	10 bench · 2 src	$1.25
99	GPT-5.4 mini	OpenAI	58th	10 bench · 2 src	$4.50
100	Qwen3-235B-A22B	Alibaba	58th	7 bench · 1 src	—
101	Gemini 2.0 Pro	Google	57th	3 bench · 1 src	$4.998
102	DeepSeek-V3.2-Exp	DeepSeek	57th	7 bench · 3 src	$0.326
103	GPT-5.6 Terra	OpenAI	56th	4 bench · 1 src	—
104	Qwen3.6 27B	Alibaba	56th	5 bench · 1 src	—
105	Qwen3.7 Flash	Alibaba	56th	5 bench · 1 src	$0.13
106	Claude Haiku 4.5	Anthropic	55th	7 bench · 1 src	$5.00
107	o1-mini	OpenAI	55th	5 bench · 2 src	$4.40
108	Claude Opus 4	Anthropic	54th	7 bench · 2 src	$75.00
109	o4-mini	OpenAI	54th	12 bench · 3 src	$4.40
110	qwen3-4b-instruct-2507	—	53rd	3 bench · 1 src	—
111	Qwen3.6 35B-A3B	Alibaba	53rd	4 bench · 1 src	$2.052
112	Claude Sonnet 4	Anthropic	53rd	7 bench · 3 src	$15.00
113	Grok 3 Beta	xAI	52nd	6 bench · 2 src	—
114	Claude Sonnet 4.5	Anthropic	52nd	10 bench · 1 src	$15.00
115	gpt-oss-20b	OpenAI	51st	3 bench · 1 src	$0.30
116	Gemini 2.0 Flash Thinking	Google	51st	6 bench · 3 src	$0.42
117	GLM 5	Zhipu	50th	6 bench · 2 src	$3.2647
118	Claude 3.7 Sonnet	Anthropic	50th	7 bench · 3 src	$15.00
119	Gemma 3 27B	Google	50th	5 bench · 2 src	free
120	Qwen3.7 Flash	Alibaba	49th	5 bench · 1 src	—
121	Inkling	Thinking Machines	49th	7 bench · 2 src	$4.05
122	Claude Opus 4.5	Anthropic	48th	11 bench · 2 src	$25.00
123	Claude Opus 4.5	Anthropic	48th	6 bench · 1 src	—
124	GLM-4.7	Zhipu	47th	6 bench · 1 src	$0.80
125	o1-2024-12-17	OpenAI	47th	9 bench · 3 src	$60.00
126	DeepSeek v4	DeepSeek	47th	4 bench · 1 src	—
127	Qwen3-14B	Alibaba	46th	3 bench · 1 src	$0.24
128	Mistral Medium 3	Mistral	46th	4 bench · 1 src	$2.00
129	DeepSeek-V3	DeepSeek	45th	4 bench · 1 src	—
130	Qwen3	Alibaba	45th	3 bench · 1 src	—
131	GPT 5 nano	OpenAI	44th	11 bench · 2 src	$0.40
132	Qwen3 32B	Alibaba	44th	4 bench · 2 src	$0.55
133	Qwen3.6 35B-A3B	Alibaba	44th	5 bench · 1 src	—
134	O3 Mini	OpenAI	43rd	12 bench · 3 src	$4.40
135	Qwen3.6 Flash	Alibaba	43rd	8 bench · 1 src	$1.16
136	Qwen3-30B-A3B	Alibaba	42nd	3 bench · 1 src	$0.50
137	Gemini 2.5 Pro	Google	42nd	8 bench · 2 src	$10.00
138	DeepSeek-R1-0528	—	42nd	3 bench · 1 src	$0.20
139	GPT-5.6 Luna	OpenAI	41st	5 bench · 1 src	—
140	Qwen3.5 9B	Alibaba	41st	3 bench · 1 src	$0.15
141	GPT 4.1 mini	OpenAI	40th	10 bench · 3 src	$1.60
142	Qwen Plus	Alibaba	40th	3 bench · 1 src	—
143	GPT-4.1 nano	OpenAI	39th	6 bench · 2 src	$0.40
144	GLM-5.2	Zhipu	39th	5 bench · 1 src	—
145	Phi-4	Microsoft	39th	4 bench · 1 src	$0.50
146	Qwen3.5 Flash	Alibaba	38th	8 bench · 1 src	$0.40
147	Llama 4 Maverick Instruct	Meta	38th	6 bench · 3 src	$0.60
148	Grok 4 Heavy	xAI	37th	1 bench · 1 src	—
149	Qwen3.6 Plus	Alibaba	37th	2 bench · 1 src	—
150	GPT-5.5 Instant	OpenAI	36th	5 bench · 1 src	$30.00
151	GPT 4.1	OpenAI	36th	10 bench · 3 src	$8.00
152	Qwen2.5-72B	Alibaba	36th	3 bench · 1 src	$0.408
153	Qwen2-72B	Alibaba	35th	2 bench · 1 src	$0.90
154	Qwen2.5-Max	Alibaba	35th	4 bench · 1 src	$6.392
155	Gemini 2.5 Flash	Google	34th	2 bench · 1 src	—
156	Gemini 1.5 Pro	Google	34th	3 bench · 1 src	—
157	DeepSeek-V3	DeepSeek	33rd	4 bench · 1 src	$0.77
158	Gemini 3.1 Flash-Lite	Google	33rd	4 bench · 1 src	$1.50
159	GPT-4 Turbo	OpenAI	33rd	2 bench · 1 src	—
160	Gemini 3.5 Flash-Lite	Google	32nd	7 bench · 2 src	$2.50
161	Claude Sonnet 4.5	Anthropic	32nd	8 bench · 1 src	—
162	Grok-2	xAI	31st	4 bench · 1 src	$10.00
163	Qwen2.5-32B	Alibaba	31st	4 bench · 1 src	$0.90
164	Claude Opus 4.1	Anthropic	30th	10 bench · 1 src	$75.00
165	Qwen Turbo	Alibaba	30th	3 bench · 1 src	$0.2006
166	GPT-5.1	OpenAI	30th	5 bench · 1 src	—
167	Mistral Small 3.2	Mistral	29th	3 bench · 1 src	$0.18
168	Llama 4 Scout Instruct	Meta	29th	5 bench · 2 src	$0.46
169	Claude 3.5 Sonnet	Anthropic	28th	5 bench · 1 src	—
170	GLM 4.6	Zhipu	28th	3 bench · 2 src	$1.40
171	Llama 3.1-405B	Meta	27th	3 bench · 1 src	—
172	Magistral Small 1.2	Mistral	27th	3 bench · 1 src	$1.50
173	Claude 3.5 Sonnet	Anthropic	27th	5 bench · 1 src	—
174	GLM 4.5	Zhipu	26th	3 bench · 2 src	$1.30
175	MiniMax M3	MiniMax	26th	4 bench · 1 src	—
176	Magistral Small 1.0	Mistral	25th	3 bench · 1 src	—
177	glm-4.7-flash_none	—	25th	3 bench · 1 src	—
178	Mistral Large 2	Mistral	24th	4 bench · 1 src	—
179	Mistral Small 3.1	Mistral	24th	4 bench · 1 src	—
180	Mistral Small 3	Mistral	24th	4 bench · 1 src	—
181	Qwen3.6 Flash	Alibaba	23rd	2 bench · 1 src	—
182	Gemini 1.5 Flash	Google	23rd	4 bench · 1 src	—
183	GPT-5.4 nano	OpenAI	22nd	5 bench · 1 src	—
184	Claude 3.5 Haiku	Anthropic	22nd	4 bench · 1 src	—
185	Qwen3 8B	Alibaba	21st	3 bench · 1 src	—
186	Tülu 3 70B	Ai2	21st	3 bench · 1 src	—
187	Llama 3.3 70B	Meta	21st	3 bench · 1 src	free
188	GPT-5.4 mini	OpenAI	20th	5 bench · 1 src	—
189	Qwen Plus	Alibaba	20th	1 bench · 1 src	—
190	DeepSeek-R1-Distill-Qwen-1.5B	DeepSeek	19th	3 bench · 1 src	free
191	WizardLM-2 8x22B	Microsoft	19th	2 bench · 1 src	$0.493
192	Yi-1.5-34B	01.AI	18th	2 bench · 1 src	—
193	Llama 3.2 90B	Meta	18th	3 bench · 1 src	$0.72
194	Claude 3 Opus	Anthropic	18th	6 bench · 2 src	$75.00
195	Mixtral 8x22B	Mistral	17th	2 bench · 1 src	$1.20
196	Llama 3.1-70B	Meta	17th	3 bench · 1 src	$0.72
197	Qwen3.5 Flash	Alibaba	16th	2 bench · 1 src	—
198	GPT 4o	OpenAI	16th	9 bench · 3 src	$10.00
199	GPT-4o mini	OpenAI	15th	8 bench · 2 src	$0.60
200	Gemma 3 12B	Google	15th	3 bench · 1 src	$0.50
201	GPT-4 Turbo	OpenAI	15th	5 bench · 1 src	—
202	Gemma 2 27B	Google	14th	3 bench · 1 src	$0.65
203	phi-3-medium 14B	Microsoft	14th	2 bench · 1 src	—
204	Llama 3-70B	Meta	13th	3 bench · 1 src	$2.75
205	Mistral Large	Mistral	13th	3 bench · 1 src	$6.001
206	Hermes 2 Theta Llama-3 70B	Nous Research	12th	3 bench · 1 src	—
207	Llama 3.1-8B	Meta	12th	4 bench · 1 src	$0.22
208	GPT-4	OpenAI	12th	5 bench · 1 src	—
209	Qwen3-1.7B	Alibaba	11th	3 bench · 1 src	$0.10
210	Ministral 8B	Mistral	11th	2 bench · 1 src	$0.15
211	Gemma 2 9B	Google	10th	3 bench · 1 src	$0.20
212	Gemma 3 4B	Google	10th	3 bench · 1 src	$0.2006
213	Ministral 3B	Mistral	9th	2 bench · 1 src	$0.10
214	Claude 3 Sonnet	Anthropic	9th	3 bench · 1 src	$15.00
215	Claude 3 Haiku	Anthropic	9th	3 bench · 1 src	$1.25
216	Gemini 1.5 Flash 8B	Google	8th	2 bench · 1 src	—
217	DBRX	Databricks	8th	2 bench · 1 src	$1.20
218	Claude 2	Anthropic	7th	4 bench · 2 src	—
219	Mistral NeMo	Mistral	7th	2 bench · 1 src	$0.17
220	granite-4.0-micro	—	6th	3 bench · 1 src	—
221	Qwen2.5-7B	Alibaba	6th	3 bench · 1 src	$0.05
222	Gemini 1.0 Pro	Google	6th	3 bench · 1 src	—
223	Claude 2.1	Anthropic	5th	2 bench · 1 src	—
224	Mixtral 8x7B	Mistral	5th	2 bench · 1 src	$0.50
225	Gemma 3 1B	Google	4th	3 bench · 1 src	—
226	granite-4.0-1b	—	4th	3 bench · 1 src	—
227	GPT-3.5 Turbo	OpenAI	3rd	6 bench · 1 src	$1.50
228	Llama 3.2 1B	Meta	3rd	3 bench · 1 src	$0.10
229	GPT-4	OpenAI	3rd	2 bench · 1 src	—
230	granite-4.0-350m	—	2nd	3 bench · 1 src	—
231	Llama 3-8B	Meta	2nd	4 bench · 1 src	$0.25
232	DeepSeek LLM 67B	DeepSeek	1st	4 bench · 1 src	—
233	Mistral 7B	Mistral	1st	4 bench · 1 src	$0.20
234	Llama 2-70B	Meta	0th	3 bench · 1 src	$2.75
235	Yi-34B	01.AI	0th	2 bench · 1 src	$0.90

Common questions

What is the best ai for math right now?: Ranked by the mathematics composite: MATH level 5, FrontierMath and OTIS Mock AIME.
How is this ranking computed?: It is the same family composite as the leaderboard. Each benchmark is weighted by how much of its spread survives its own measurement error.