How is this ranking computed?

The set is a spec filter on the catalog. Scored models rank by the overall composite. Unscored models follow by the spec itself.

Best open-weight AI models

Models whose listings report open weights. Ranked by composite when scored, else by output price. Full leaderboard.

Rank	Model	Lab	Composite	Context	Out $/M
1	DeepSeek V4.1 Flash	DeepSeek	94th	1M	$1.2778
2	MiniMax-M2.5	MiniMax	91st	205K	$1.20
3	GLM-5.3	Zhipu	82nd	1M	$4.40
4	DeepSeek V4 Flash Vision Exp	DeepSeek	80th	1M	$1.32
5	DeepSeek-R1	DeepSeek	78th	128K	$1.70
6	GLM-5.1	Zhipu	77th	200K	$4.40
7	Qwen3.6 35B-A3B	Alibaba	73rd	262K	$2.052
8	Kimi K2	Kimi	72nd	128K	free
9	Kimi K2.6	Kimi	72nd	262K	$4.00
10	Moonshotai/Kimi-K2.5	Alibaba	71st	256K	$1.90
11	Qwen3.5 397B-A17B	Alibaba	70th	262K	$2.65
12	DeepSeek R1 Distill Llama 70B	DeepSeek	69th	33K	$0.99
13	DeepSeek V3.2	DeepSeek	68th	128K	$0.326
14	Minimax/Minimax-M2	MiniMax	68th	200K	$1.53
15	GLM-5	Alibaba	66th	205K	$3.2647
16	GLM-5.2	Alibaba	66th	1M	$4.40
17	Kimi K2.7 Code	Kimi	63rd	262K	$4.389
18	Qwen3.5 35B A3B	Alibaba	63rd	260K	$1.80
19	GLM-5.3-Flash	Zhipu	61st	1M	$0.025
20	Inkling Thinking	Databricks	61st	1M	$4.05
21	Z-Ai/GLM 4.7	Zhipu	51st	200K	$0.80
22	Qwen3.5 9B	Alibaba	48th	256K	$0.15
23	GPT OSS 120B	OpenAI	47th	131K	$0.798
24	Qwen3.5 122B A10B Thinking	Alibaba	45th	131K	$3.496
25	GLM 4.7 Flash Thinking	Zhipu	43rd	200K	$0.40
26	Z-AI/GLM 4.6	Zhipu	38th	200K	$1.40
27	DeepSeek-V3	DeepSeek	35th	128K	$0.77
28	DeepSeek V4 Pro	DeepSeek	35th	1M	$0.87
29	GLM 4.5	Zhipu	34th	131K	$1.30
30	Mistral Large 2411	Mistral	15th	128K	$6.001
31	Cerebras-Llama-4-Scout-17B-16E-Instruct	Meta	—	128K	free
32	Gemma 4 31B IT (free)	Google	—	262K	free
33	MiMo-V2.5-Pro	—	—	1M	free
34	Llama 3.2 3b Instruct	Meta	—	131K	$0.0493
35	Llama 3.1 8b Instruct	Meta	—	131K	$0.085
36	gpt-oss-20b	OpenAI	—	128K	$0.15
37	DeepSeek V4 Flash	DeepSeek	—	1M	$0.196
38	Tencent Hy3	—	—	262K	$0.26
39	Mistral Small 3.2 (Mistral AI)	Mistral	—	128K	$0.30
40	Qwen3 30B A3B	Alibaba	—	40K	$0.30
41	Qwen3 32B	Alibaba	—	40K	$0.30
42	Stepfun/Step-3.5 Flash	—	—	64K	$0.30
43	Gemma 4 26B A4B	Google	—	262K	$0.38
44	DeepSeek V3.2 Exp	DeepSeek	—	164K	$0.42
45	Qwen 3 235B A22B	Alibaba	—	128K	$0.50
46	Qwen3 235B A22B Thinking 2507	Alibaba	—	262K	$0.60
47	Qwen3 Next 80B A3B Instruct	Alibaba	—	131K	$0.65
48	Qwen3 Next 80B A3B Thinking	Alibaba	—	131K	$0.65
49	Gemma 3 27B	Google	—	40K	$0.684
50	DeepSeek-V3.1	DeepSeek	—	128K	$0.70
51	Qwen3.8 27B	Alibaba	—	262K	$0.70
52	GLM 4.5 Air	Zhipu	—	131K	$0.80
53	Codestral 2508	Mistral	—	256K	$0.90
54	Qwen3 Coder 480B A35B Instruct	Alibaba	—	262K	$1.00
55	Qwen3-Coder 30B-A3B Instruct	Alibaba	—	128K	$1.083
56	Step 3.7 Flash	—	—	262K	$1.14
57	MiniMax M3 Thinking	MiniMax	—	512K	$1.20
58	Qwen3 VL 235B A22B Instruct	Alibaba	—	131K	$1.20
59	Llama-3.3-70B-Instruct	Meta	—	100K	$1.254
60	MiniMax M2.7	MiniMax	—	205K	$1.26
61	Minimax/Minimax-M2.1	MiniMax	—	205K	$1.32
62	Llama 4 Maverick 17B 128E Instruct FP8	Meta	—	131K	$1.484
63	Qwen3 Coder Flash	Alibaba	—	128K	$1.50
64	Qwen3 Coder Next	Alibaba	—	262K	$1.50
65	GLM 4.5V Thinking	Zhipu	—	66K	$1.80
66	Qwen 3.6 Plus	Alibaba	—	992K	$1.95
67	MiMo-V2.5	—	—	1M	$2.00
68	Qwen3.5 27B	Alibaba	—	260K	$2.16
69	Qwen3.6 27B Thinking	Alibaba	—	260K	$2.24
70	MiniMax-M2.7-highspeed	MiniMax	—	205K	$2.40
71	Kimi K2 Thinking	Kimi	—	256K	$2.50
72	Qwen3 235B A22B Instruct 2507	Alibaba	—	262K	$3.078
73	Qwen3 Coder Plus	Alibaba	—	128K	$5.00
74	Qwen3 VL 235B A22B Thinking	Alibaba	—	131K	$6.00
75	Kimi K3	Alibaba	—	1M	$15.00

Common questions

What is the best open-weight ai models right now?: Models whose listings report open weights. Ranked by composite when scored, else by output price.
How is this ranking computed?: The set is a spec filter on the catalog. Scored models rank by the overall composite. Unscored models follow by the spec itself.