Ranked by composite score

AI model leaderboard

One number per model, built from 16 published benchmarks across 4 sources. Each benchmark is weighted by how much of its spread survives its own measurement error, so a benchmark that no longer separates models stops counting. How the weights are computed. The quality rank uses scores only. Best per dollar is quality over output price. Speed is not in any rank. We do not sample live APIs.

79 models ranked · 3 or more benchmarks, from 2 or more sources · percentile is out of all 334 scored models either way

Rank	Model	Lab	Composite	Evidence	Effort range	Out $/M
1	GPT-6 Astra	OpenAI	100th	11 bench · 2 src	9.8 pts	$50.00
2	Claude Fable 5	Anthropic	99th	14 bench · 2 src	7.1 pts	$50.00
3	Claude Fable 5.1	Anthropic	99th	10 bench · 2 src	—	$50.00
4	Claude Opus 5	Anthropic	96th	9 bench · 2 src	14.0 pts	$25.00
5	Gemini 3.8 Flash	Google	95th	8 bench · 2 src	—	$3.75
6	GPT-5.6 Sol	OpenAI	94th	12 bench · 2 src	16.2 pts	$10.00
7	Gemini 3.7 Flash	Google	94th	8 bench · 2 src	—	$3.75
8	GPT-5.5	OpenAI	93rd	16 bench · 2 src	21.8 pts	$30.00
9	DeepSeek V4 Pro 0813	DeepSeek	91st	8 bench · 2 src	—	$4.0077
10	Gemini 3 Pro Preview	Google	91st	6 bench · 2 src	6.2 pts	$9.60
11	GPT-5.6 Terra	OpenAI	90th	8 bench · 2 src	17.9 pts	$12.00
12	Qwen3.8 Max	Alibaba	89th	8 bench · 2 src	—	$6.00
13	Claude Opus 4.8	Anthropic	89th	13 bench · 2 src	3.8 pts	$25.00
14	GPT-5.4	OpenAI	88th	13 bench · 2 src	16.7 pts	$15.00
15	Grok 4.6	xAI	87th	9 bench · 2 src	1.1 pts	$6.00
16	Gemini 3.1 Pro Preview	Google	86th	12 bench · 2 src	2.7 pts	$12.00
17	Grok 4.5	xAI	85th	7 bench · 2 src	—	$6.00
18	Kimi K3	Kimi	85th	8 bench · 2 src	16.5 pts	free
19	Gemini 3 Flash Preview	Google	84th	10 bench · 2 src	2.8 pts	$3.00
20	Gemini 3.6 Flash	Google	84th	8 bench · 2 src	8.1 pts	$3.75
21	Gemini 3.5 Flash	Google	84th	14 bench · 3 src	7.0 pts	$9.00
22	Claude Opus 4.7	Anthropic	83rd	13 bench · 2 src	11.1 pts	$25.00
23	GLM-5.3	Zhipu	82nd	8 bench · 2 src	—	$4.40
24	Grok 4	xAI	82nd	6 bench · 2 src	—	$15.00
25	DeepSeek V4 Flash 0731	DeepSeek	81st	8 bench · 2 src	—	$0.50
26	Claude Sonnet 5	Anthropic	80th	8 bench · 2 src	12.5 pts	$10.00
27	GPT-5.6 Luna	OpenAI	80th	8 bench · 2 src	19.0 pts	$0.37
28	GPT 5.2	OpenAI	79th	12 bench · 3 src	14.3 pts	$14.00
29	Qwen3.7-Max	Alibaba	79th	10 bench · 2 src	—	$9.00
30	DeepSeek R1	DeepSeek	78th	4 bench · 2 src	—	$1.70
31	GPT 5	OpenAI	78th	13 bench · 3 src	8.8 pts	$10.00
32	o3	OpenAI	76th	11 bench · 3 src	4.2 pts	$8.00
33	GPT 5.1	OpenAI	75th	8 bench · 2 src	8.3 pts	$10.00
34	Claude Opus 4.6	Anthropic	74th	11 bench · 2 src	4.3 pts	$25.00
35	Kimi K2 Instruct	Kimi	72nd	7 bench · 3 src	—	free
36	Kimi K2.6	Kimi	72nd	12 bench · 2 src	—	$4.00
37	Kimi K2.5	Kimi	71st	7 bench · 2 src	3.0 pts	$1.90
38	Claude Opus 4	Anthropic	70th	7 bench · 2 src	4.6 pts	$75.00
39	DeepSeek-V3.2-Exp	DeepSeek	68th	7 bench · 3 src	10.0 pts	$0.326
40	Qwen3.6 Plus	Alibaba	67th	9 bench · 2 src	—	$3.00
41	Claude Sonnet 4	Anthropic	66th	7 bench · 3 src	11.6 pts	$15.00
42	GLM 5	Zhipu	66th	6 bench · 2 src	0.7 pts	$3.2647
43	GLM-5.2	Zhipu	66th	10 bench · 2 src	4.5 pts	$4.40
44	Claude Sonnet 4.6	Anthropic	64th	8 bench · 2 src	9.3 pts	$15.94
45	Claude 3.7 Sonnet	Anthropic	64th	7 bench · 3 src	12.5 pts	$15.00
46	Kimi K2.7 Code	Kimi	63rd	7 bench · 2 src	—	$4.389
47	GPT-4.5	OpenAI	62nd	4 bench · 2 src	—	$150.00
48	Qwen3.6 27B	Alibaba	62nd	5 bench · 2 src	—	free
49	GLM-5.3-Flash	Zhipu	61st	7 bench · 2 src	—	$0.025
50	Inkling	Thinking Machines	61st	7 bench · 2 src	—	$4.05
51	Grok 3 Betanot in catalog	xAI	59th	6 bench · 2 src	—	—
52	o1-2024-12-17	OpenAI	59th	9 bench · 3 src	—	$60.00
53	Gemini 2.5 Pro	Google	57th	8 bench · 2 src	4.0 pts	$10.00
54	Claude Opus 4.5	Anthropic	55th	11 bench · 2 src	3.3 pts	$25.00
55	o4-mini	OpenAI	55th	12 bench · 3 src	—	$4.40
56	GPT 5 mini	OpenAI	54th	11 bench · 2 src	3.3 pts	$2.00
57	MiniMax M3	MiniMax	53rd	5 bench · 2 src	—	$2.40
58	GPT-5.4 mini	OpenAI	50th	10 bench · 2 src	3.3 pts	$4.50
59	o1-mini	OpenAI	49th	5 bench · 2 src	—	$4.40
60	gpt-oss-120b	OpenAI	47th	6 bench · 3 src	—	$0.798
61	GPT-5.4 nano	OpenAI	46th	10 bench · 2 src	13.0 pts	$1.25
62	O3 Mini	OpenAI	42nd	12 bench · 3 src	6.6 pts	$4.40
63	GPT 4.1	OpenAI	41st	10 bench · 3 src	9.0 pts	$8.00
64	Qwen3 32B	Alibaba	41st	4 bench · 2 src	—	$0.55
65	QwQ 32B	Alibaba	41st	4 bench · 2 src	—	$0.861
66	GPT 5 nano	OpenAI	40th	11 bench · 2 src	10.8 pts	$0.40
67	Gemini 3.5 Flash-Lite	Google	39th	7 bench · 2 src	8.0 pts	$2.50
68	GLM 4.6	Zhipu	38th	3 bench · 2 src	—	$1.40
69	Gemini 2.0 Flash Thinking	Google	37th	6 bench · 3 src	—	$0.42
70	GLM 4.5	Zhipu	34th	3 bench · 2 src	—	$1.30
71	GPT 4.1 mini	OpenAI	31st	10 bench · 3 src	—	$1.60
72	Llama 4 Maverick Instruct	Meta	29th	6 bench · 3 src	—	$0.60
73	Gemma 3 27B	Google	28th	5 bench · 2 src	—	free
74	GPT-4.1 nano	OpenAI	22nd	6 bench · 2 src	—	$0.40
75	GPT 4o	OpenAI	21st	9 bench · 3 src	2.0 pts	$10.00
76	Llama 4 Scout Instruct	Meta	20th	5 bench · 2 src	—	$0.46
77	GPT-4o mini	OpenAI	15th	8 bench · 2 src	—	$0.60
78	Claude 3 Opus	Anthropic	14th	6 bench · 2 src	—	$75.00
79	Claude 2not in catalog	Anthropic	2nd	4 bench · 2 src	—	—

Effort range is the median spread between the same model's reasoning-effort settings, in points. It is larger than the disagreement between sources, so a score quoted without its configuration is under-specified.

Evidence counts the benchmarks behind the number and the independent sources behind those. The percentile denominator never changes with the filter.

Open weights uses the catalog flag. Closed is every catalog model without that flag. A scored model with no catalog row is neither.

Paid is a listed output price above zero. Free is a listed output price of zero. An open-weight model can have a paid API.