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 reasoning

Ranked by the reasoning composite: GPQA Diamond, EBR-bench, Chess Puzzles and Mystery Game Puzzles. Full leaderboard.

Rank	Model	Lab	Composite	Evidence	Out $/M
1	GPT-6 Astra	OpenAI	100th	1 bench · 1 src	$50.00
2	GPT-6 Astra	OpenAI	100th	15 bench · 2 src	$50.00
3	GPT-5.6 Sol	OpenAI	99th	2 bench · 1 src	$30.00
4	Claude Opus 5.5	Anthropic	99th	15 bench · 2 src	$20.00
5	GPT-6.1 Sol	OpenAI	98th	13 bench · 2 src	$10.00
6	Claude Sonnet 5.5	Anthropic	98th	13 bench · 2 src	$10.00
7	Gemini 4 Argon	Google	98th	9 bench · 1 src	$20.00
8	GPT-5.5 Pro	OpenAI	97th	3 bench · 1 src	$180.00
9	GPT-5.2 Pro	OpenAI	97th	3 bench · 1 src	$168.00
10	Claude Fable 5.1	Anthropic	96th	14 bench · 2 src	$50.00
11	GPT-5.4 Pro	OpenAI	96th	6 bench · 1 src	$180.00
12	GPT-6 Sol	OpenAI	96th	13 bench · 2 src	$10.00
13	Gemini 3.8 Flash	Google	95th	9 bench · 2 src	$7.50
14	Gemini 3.7 Flash	Google	95th	9 bench · 2 src	$7.50
15	GPT-5.5 Pro pre-release	OpenAI	94th	4 bench · 1 src	—
16	Gemini 3 Pro Preview	Google	94th	5 bench · 2 src	$12.00
17	DeepSeek V4.1 Flash	DeepSeek	94th	9 bench · 2 src	$1.20
18	DeepSeek V4 Pro 0813	DeepSeek	93rd	8 bench · 2 src	$3.96
19	Muse Spark	Meta	93rd	4 bench · 1 src	—
20	Muse Spark Open	Meta	92nd	5 bench · 1 src	$4.25
21	GPT-5 Pro	OpenAI	92nd	5 bench · 1 src	$120.00
22	GPT-5.5	OpenAI	92nd	17 bench · 2 src	$30.00
23	Qwen3.8 Max	Alibaba	91st	8 bench · 2 src	$6.00
24	GLM 5	Zhipu	91st	5 bench · 2 src	$3.20
25	Kimi K2.5	Kimi	90th	6 bench · 2 src	$2.80
26	Qwen3.6 Max Preview	Alibaba	90th	6 bench · 1 src	free
27	Grok 4	xAI	90th	5 bench · 2 src	$15.00
28	GPT-5.6 Sol	OpenAI	89th	13 bench · 2 src	$20.00
29	Claude Fable 5	Anthropic	89th	15 bench · 2 src	$50.00
30	Claude Opus 5	Anthropic	88th	10 bench · 2 src	$25.00
31	Qwen3.6 27B	Alibaba	88th	4 bench · 2 src	$2.4754
32	Qwen3.5 397B-A17B	Alibaba	88th	3 bench · 1 src	$1.032
33	Gemini 2.5 Pro	Google	87th	7 bench · 2 src	$10.00
34	Muse Wonder	Meta	87th	6 bench · 1 src	$20.00
35	Kimi K2 Instruct	Kimi	86th	6 bench · 3 src	$2.5146
36	Qwen3.8 Max (0902)	Alibaba	86th	7 bench · 1 src	—
37	Qwen3.6 35B-A3B	Alibaba	86th	3 bench · 1 src	free
38	Gemini 2.5 Pro Exp	Google	85th	2 bench · 1 src	—
39	Grok 4.5	xAI	85th	8 bench · 2 src	$6.00
40	GPT-5.5 Instant	OpenAI	84th	4 bench · 1 src	—
41	Qwen3.5-35B-A3B	Alibaba	84th	2 bench · 1 src	$0.459
42	Grok 4.7	xAI	84th	9 bench · 2 src	$6.00
43	GLM-5.3	Zhipu	83rd	8 bench · 2 src	$4.40
44	Claude Sonnet 5	Anthropic	83rd	8 bench · 2 src	$10.00
45	Claude Haiku 5.5	Anthropic	82nd	10 bench · 2 src	$0.50
46	DeepSeek V4 Flash 0731	DeepSeek	82nd	8 bench · 2 src	$0.28
47	Qwen3.5 9B	Alibaba	82nd	2 bench · 1 src	$0.13
48	Qwen3.5 9B	Alibaba	81st	2 bench · 1 src	—
49	Claude Opus 4.8	Anthropic	81st	14 bench · 2 src	$25.00
50	Grok 4.20	xAI	80th	6 bench · 1 src	$12.00
51	Gemini 3.1 Pro Preview	Google	80th	13 bench · 2 src	$12.00
52	Gemini 3.6 Flash	Google	80th	9 bench · 2 src	$3.75
53	Grok 4.6	xAI	79th	10 bench · 2 src	$6.00
54	Gemini 3 Flash Preview	Google	79th	10 bench · 2 src	$3.00
55	DeepSeek-V3.2-Exp	DeepSeek	78th	6 bench · 3 src	$0.89
56	Claude Sonnet 4	Anthropic	78th	7 bench · 3 src	—
57	Claude 3.7 Sonnet	Anthropic	78th	7 bench · 3 src	$15.00
58	DeepSeek-R1	DeepSeek	77th	4 bench · 1 src	—
59	Gemma 4 31B IT	Google	77th	3 bench · 1 src	free
60	Gemini 2.5 Pro Preview	Google	76th	4 bench · 1 src	—
61	Kimi K3	Kimi	76th	9 bench · 2 src	$1.00
62	Grok-3 mini	xAI	76th	5 bench · 1 src	$0.50
63	GPT-5.6 Terra	OpenAI	75th	9 bench · 2 src	$12.00
64	GPT-5.4	OpenAI	75th	14 bench · 2 src	$15.00
65	o3	OpenAI	74th	11 bench · 3 src	$8.00
66	Qwen3.6 35B-A3B	Alibaba	74th	4 bench · 1 src	—
67	Gemini 3.1 Flash-Lite	Google	74th	3 bench · 1 src	$1.50
68	Nemotron 3 Ultra	NVIDIA	73rd	3 bench · 1 src	$0.10
69	Gemma 4 26B-A4B IT	Google	73rd	2 bench · 1 src	$0.29
70	GLM-5.1	Zhipu	72nd	8 bench · 1 src	$4.40
71	Inkling	Thinking Machines	72nd	7 bench · 2 src	$4.05
72	Claude Opus 4	Anthropic	72nd	7 bench · 2 src	—
73	Qwen3.7 Plus	Alibaba	71st	3 bench · 1 src	$1.60
74	Qwen3.5 397B-A17B	Alibaba	71st	4 bench · 1 src	—
75	GPT 5.2	OpenAI	70th	13 bench · 3 src	$14.00
76	Mistral Large 4	Mistral	70th	9 bench · 2 src	$4.4156
77	Muse Spark 1.1	Meta	70th	3 bench · 2 src	$4.25
78	Claude Opus 4.7	Anthropic	69th	14 bench · 2 src	$25.00
79	DeepSeek v4	DeepSeek	69th	7 bench · 1 src	—
80	seed-oss-36b-instruct	—	68th	2 bench · 1 src	$0.5978
81	Grok 4.3 Beta	xAI	68th	6 bench · 1 src	—
82	GPT 5.1	OpenAI	68th	8 bench · 2 src	$10.00
83	Qwen3	Alibaba	67th	3 bench · 1 src	—
84	Gemini 3.5 Flash	Google	67th	14 bench · 3 src	$9.00
85	GPT-5.6 Luna	OpenAI	66th	9 bench · 2 src	$1.20
86	GLM-4.7	Zhipu	66th	6 bench · 1 src	$2.20
87	Claude Opus 4.8	Anthropic	66th	3 bench · 1 src	—
88	GPT-6 Luna	OpenAI	65th	12 bench · 2 src	$0.50
89	Qwen3.7 Plus	Alibaba	65th	4 bench · 1 src	—
90	DeepSeek R1	DeepSeek	64th	4 bench · 2 src	$2.60
91	Qwen3.6 Plus	Alibaba	64th	9 bench · 2 src	$3.00
92	GPT-4.5	OpenAI	64th	4 bench · 2 src	$150.00
93	Kimi K2.7 Code	Kimi	63rd	7 bench · 2 src	$4.00
94	Qwen3.7 Flash	Alibaba	63rd	4 bench · 1 src	$0.13
95	DeepSeek-V3	DeepSeek	62nd	4 bench · 1 src	—
96	Grok 3 Beta	xAI	62nd	6 bench · 2 src	—
97	Qwen3.7-Max	Alibaba	61st	10 bench · 2 src	$7.50
98	GPT-5.6 Sol	OpenAI	61st	4 bench · 1 src	—
99	Qwen3.7 Flash	Alibaba	61st	4 bench · 1 src	—
100	Muse Spark 1.3	Meta	60th	6 bench · 2 src	$4.25
101	Llama 4 Maverick Instruct	Meta	60th	6 bench · 3 src	$0.60
102	Inkling Small	Thinking Machines	59th	6 bench · 1 src	$1.20
103	Gemini 2.0 Pro	Google	59th	3 bench · 1 src	$4.998
104	QwQ-Plus	Alibaba	59th	1 bench · 1 src	$2.40
105	QwQ 32B	Alibaba	58th	3 bench · 2 src	$1.00
106	Claude Sonnet 4.6	Anthropic	58th	8 bench · 2 src	$15.94
107	Qwen3.6 27B	Alibaba	57th	4 bench · 1 src	—
108	GLM-5.3-Flash	Zhipu	57th	7 bench · 2 src	$0.50
109	Qwen3.5 Flash	Alibaba	57th	8 bench · 1 src	$0.368
110	DeepSeek-R1-Distill-Qwen-32B	DeepSeek	56th	2 bench · 1 src	$0.861
111	Muse Spark 1.2	Meta	56th	3 bench · 2 src	$4.25
112	GPT-5.4 nano	OpenAI	55th	10 bench · 2 src	$1.25
113	Qwen3-235B-A22B	Alibaba	55th	6 bench · 1 src	—
114	GPT 5	OpenAI	55th	13 bench · 3 src	$10.00
115	Qwen3-30B-A3B	Alibaba	54th	2 bench · 1 src	—
116	Claude Opus 4.6	Anthropic	54th	12 bench · 2 src	$25.00
117	GPT-5.4 mini	OpenAI	53rd	10 bench · 2 src	$4.50
118	GLM-5.2	Zhipu	53rd	10 bench · 2 src	$4.40
119	GLM-5.2	Zhipu	53rd	5 bench · 1 src	—
120	Claude Sonnet 4.5	Anthropic	52nd	10 bench · 1 src	$15.00
121	Qwen3.5 Plus	Alibaba	52nd	7 bench · 1 src	$2.21
122	MiniMax M3	MiniMax	51st	5 bench · 2 src	$1.20
123	o1-mini	OpenAI	51st	5 bench · 2 src	$4.40
124	Gemini 3.5 Flash-Lite	Google	51st	7 bench · 2 src	$2.50
125	Gemini 2.0 Flash Thinking	Google	50th	6 bench · 3 src	$0.42
126	glm-4.7-flash	—	50th	2 bench · 1 src	$0.40
127	Claude Opus 4.5	Anthropic	49th	12 bench · 2 src	$25.00
128	Qwen3.6 Flash	Alibaba	49th	8 bench · 1 src	$1.50
129	Claude Opus 4.1	Anthropic	49th	9 bench · 1 src	$75.00
130	Qwen3 32B	Alibaba	48th	3 bench · 2 src	—
131	Claude Opus 4.5	Anthropic	48th	6 bench · 1 src	—
132	GPT-5.5	OpenAI	47th	4 bench · 1 src	—
133	Qwen3.6 Max Preview	Alibaba	47th	1 bench · 1 src	—
134	Mistral Medium 3	Mistral	47th	4 bench · 1 src	$7.50
135	Kimi K2.6	Kimi	46th	13 bench · 2 src	$4.00
136	Qwen3-14B	Alibaba	46th	2 bench · 1 src	$0.24
137	Qwen3.6 Flash	Alibaba	45th	2 bench · 1 src	—
138	o4-mini	OpenAI	45th	12 bench · 3 src	$4.00
139	Qwen3 Max	Alibaba	45th	6 bench · 1 src	$6.00
140	Qwen3.5 Plus	Alibaba	44th	1 bench · 1 src	—
141	Qwen3.5-122B-A10B	Alibaba	44th	1 bench · 1 src	$0.917
142	Qwen3 8B	Alibaba	43rd	2 bench · 1 src	free
143	DeepSeek-V3	DeepSeek	43rd	4 bench · 1 src	$1.10
144	GPT-5.6 Terra	OpenAI	43rd	4 bench · 1 src	—
145	o1-2024-12-17	OpenAI	42nd	9 bench · 3 src	$54.00
146	Qwen2.5-Max	Alibaba	42nd	4 bench · 1 src	—
147	Phi-4	Microsoft	41st	3 bench · 1 src	$0.50
148	Magistral Small 1.0	Mistral	41st	3 bench · 1 src	—
149	Qwen3-30B-A3B	Alibaba	41st	2 bench · 1 src	$0.50
150	DeepSeek v4	DeepSeek	40th	4 bench · 1 src	—
151	gpt-oss-120b	OpenAI	40th	5 bench · 3 src	free
152	MiniMax M3	MiniMax	39th	4 bench · 1 src	—
153	DeepSeek-R1-Distill-Llama-70B	DeepSeek	39th	3 bench · 1 src	$0.13
154	GPT 5 mini	OpenAI	39th	11 bench · 2 src	$2.00
155	GPT-5.4	OpenAI	38th	4 bench · 1 src	—
156	Qwen3.5 Flash	Alibaba	38th	2 bench · 1 src	—
157	Claude Sonnet 5	Anthropic	37th	1 bench · 1 src	—
158	Claude 3.5 Sonnet	Anthropic	37th	5 bench · 1 src	—
159	GPT 5 nano	OpenAI	37th	11 bench · 2 src	$0.40
160	Claude Opus 4.6	Anthropic	36th	4 bench · 1 src	—
161	Claude 3.5 Sonnet	Anthropic	36th	5 bench · 1 src	—
162	O3 Mini	OpenAI	35th	12 bench · 3 src	$4.40
163	GPT-5.2	OpenAI	35th	4 bench · 1 src	—
164	Grok-2	xAI	35th	4 bench · 1 src	$10.00
165	GPT-5.6 Luna	OpenAI	34th	5 bench · 1 src	—
166	gpt-oss-20b	OpenAI	34th	2 bench · 1 src	$0.30
167	Claude Haiku 4.5	Anthropic	33rd	7 bench · 1 src	$5.00
168	GPT-5.1	OpenAI	33rd	5 bench · 1 src	—
169	Claude Sonnet 4.6	Anthropic	33rd	2 bench · 1 src	—
170	Llama 4 Scout Instruct	Meta	32nd	5 bench · 2 src	$0.30
171	GPT 4.1	OpenAI	32nd	10 bench · 3 src	$8.00
172	Gemini 1.5 Pro	Google	31st	3 bench · 1 src	—
173	Claude Opus 4.7	Anthropic	31st	1 bench · 1 src	—
174	Llama 3.1-405B	Meta	31st	3 bench · 1 src	—
175	Mistral Large 2	Mistral	30th	4 bench · 1 src	—
176	Qwen3.6 Plus	Alibaba	30th	2 bench · 1 src	—
177	Claude Sonnet 4.5	Anthropic	29th	8 bench · 1 src	—
178	GPT-5.4 mini	OpenAI	29th	5 bench · 1 src	—
179	Qwen2.5-72B	Alibaba	29th	3 bench · 1 src	$0.40
180	Mistral Small 3.2	Mistral	28th	2 bench · 1 src	—
181	Qwen Plus	Alibaba	28th	3 bench · 1 src	—
182	Qwen3-4B	Alibaba	27th	1 bench · 1 src	$0.24
183	GPT 4.1 mini	OpenAI	27th	10 bench · 3 src	$1.60
184	Gemma 3 27B	Google	27th	4 bench · 2 src	free
185	Magistral Small 1.2	Mistral	26th	2 bench · 1 src	—
186	Qwen3 8B	Alibaba	26th	2 bench · 1 src	—
187	Llama 3.3 70B	Meta	25th	3 bench · 1 src	$0.72
188	GPT-4 Turbo	OpenAI	25th	4 bench · 1 src	—
189	Mistral Small 3	Mistral	25th	3 bench · 1 src	$0.08
190	Tülu 3 70B	Ai2	24th	3 bench · 1 src	—
191	Qwen2.5-32B	Alibaba	24th	3 bench · 1 src	$0.90
192	qwen3-4b-instruct-2507	—	23rd	2 bench · 1 src	—
193	GPT-5.4 nano	OpenAI	23rd	5 bench · 1 src	—
194	glm-4.7-flash_none	—	23rd	2 bench · 1 src	—
195	DeepSeek-R1-Distill-Qwen-14B	DeepSeek	22nd	3 bench · 1 src	$0.431
196	Mistral Small 3.1	Mistral	22nd	3 bench · 1 src	—
197	Llama 3.1-70B	Meta	21st	3 bench · 1 src	$0.72
198	GPT 4o	OpenAI	21st	9 bench · 3 src	$10.00
199	Gemini 1.5 Flash	Google	21st	4 bench · 1 src	—
200	WizardLM-2 8x22B	Microsoft	20th	2 bench · 1 src	$0.48
201	GPT-4 Turbo	OpenAI	20th	2 bench · 1 src	—
202	Qwen Turbo	Alibaba	19th	3 bench · 1 src	$0.09
203	Claude 3 Opus	Anthropic	19th	6 bench · 2 src	$75.00
204	GPT-4o mini	OpenAI	19th	8 bench · 2 src	$0.60
205	Llama 3.2 90B	Meta	18th	3 bench · 1 src	$0.72
206	GPT-4.1 nano	OpenAI	18th	7 bench · 2 src	$0.40
207	Qwen2-72B	Alibaba	17th	2 bench · 1 src	$0.90
208	Claude 3 Sonnet	Anthropic	17th	3 bench · 1 src	$15.00
209	Llama 3-70B	Meta	17th	3 bench · 1 src	$2.75
210	GPT-4	OpenAI	16th	4 bench · 1 src	—
211	Gemma 3 12B	Google	16th	2 bench · 1 src	$0.50
212	Mistral Large	Mistral	15th	3 bench · 1 src	$1.50
213	Claude 3.5 Haiku	Anthropic	15th	4 bench · 1 src	—
214	Hermes 2 Theta Llama-3 70B	Nous Research	15th	3 bench · 1 src	—
215	Gemma 2 27B	Google	14th	3 bench · 1 src	$0.65
216	Claude 3 Haiku	Anthropic	14th	3 bench · 1 src	$1.25
217	GPT-4	OpenAI	13th	2 bench · 1 src	—
218	Qwen2.5-7B	Alibaba	13th	2 bench · 1 src	$0.10
219	Claude 2	Anthropic	12th	4 bench · 2 src	—
220	Qwen3-1.7B	Alibaba	12th	2 bench · 1 src	$0.10
221	Mixtral 8x22B	Mistral	12th	2 bench · 1 src	$1.20
222	Gemini 1.0 Pro	Google	11th	3 bench · 1 src	—
223	Eurus-2-7B-PRIME	Tsinghua University	11th	1 bench · 1 src	—
224	DeepSeek-R1-Distill-Qwen-1.5B	DeepSeek	10th	2 bench · 1 src	free
225	Gemini 1.5 Flash 8B	Google	10th	2 bench · 1 src	—
226	Claude 2.1	Anthropic	10th	2 bench · 1 src	—
227	DBRX	Databricks	9th	2 bench · 1 src	$1.20
228	GPT-3.5 Turbo	OpenAI	9th	5 bench · 1 src	$1.40
229	Yi-1.5-34B	01.AI	8th	2 bench · 1 src	—
230	Qwen1.5-32B	Alibaba	8th	1 bench · 1 src	—
231	Mixtral 8x7B	Mistral	8th	2 bench · 1 src	$0.50
232	Mistral NeMo	Mistral	7th	2 bench · 1 src	$40.00
233	Qwen1.5-72B	Alibaba	7th	1 bench · 1 src	$0.90
234	granite-4.0-micro	—	6th	2 bench · 1 src	—
235	phi-3-medium 14B	Microsoft	6th	2 bench · 1 src	—
236	Gemma 2 9B	Google	6th	3 bench · 1 src	$0.20
237	Ministral 8B	Mistral	5th	2 bench · 1 src	$0.15
238	Llama 3.1-8B	Meta	5th	3 bench · 1 src	$0.22
239	Llama 2-70B	Meta	4th	3 bench · 1 src	$2.75
240	Llama 3-8B	Meta	4th	3 bench · 1 src	$0.25
241	Ministral 3B	Mistral	4th	2 bench · 1 src	$0.10
242	DeepSeek LLM 67B	DeepSeek	3rd	3 bench · 1 src	—
243	granite-4.0-1b	—	3rd	2 bench · 1 src	—
244	Llama 3.2 1B	Meta	2nd	2 bench · 1 src	$0.10
245	Gemma 3 4B	Google	2nd	2 bench · 1 src	$0.10
246	Gemma 3 1B	Google	2nd	2 bench · 1 src	—
247	Yi-34B	01.AI	1st	2 bench · 1 src	$0.90
248	Mistral 7B	Mistral	1st	3 bench · 1 src	$0.20
249	granite-4.0-350m	—	0th	2 bench · 1 src	—
250	DeepSeek-R1-0528	—	0th	2 bench · 1 src	$0.20

Common questions

What is the best ai for reasoning right now?: Ranked by the reasoning composite: GPQA Diamond, EBR-bench, Chess Puzzles and Mystery Game Puzzles.
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.