Best value

The best score for the money.

387 measured results across 3 benchmarks. 91 also have a published price. A dash means that number was never published, not that it is zero.

benchmark

GPQA

Ranked by published score, highest first.

#	Model	Score	Price / 1k tokens	Response time
1	Darwin-180B-RSI	94.4	-	-
2	Kimi-K3	93.5	-	-
3	MoonshotAI: Kimi K3	93.5	$0.01	-
4	Darwin-397B-ZTC	93.4	-	-
5	Ornith-1.5-397B	92.8	-	-
6	Qwen3.8-2.4T-A95B	92.6	-	-
7	Qwen: Qwen3.8 2.4T A95B	92.6	<$0.01	-
8	Hy4-preview	92.3	-	-
9	Tencent: Hy4 preview	92.3	<$0.01	-
10	Qwen3.8-Flash-Next	91.7	-	-
11	GLM-5.2	91.2	-	-
12	Darwin-398B-JGOS	90.9	-	-
13	DeepSeek-V4.1-Flash	90.9	-	-
14	Kimi-K2.6	90.5	-	-
15	MoonshotAI: Kimi K2.6	90.5	<$0.01	-
16	Hy3	90.4	-	-
17	Tencent: Hy3	90.4	<$0.01	-
18	DeepSeek-V4-Pro	90.1	-	-
19	Thinking Machines: Inkling Small	89.5	<$0.01	-
20	Inkling-Small	89.5	-	-
21	Darwin-28B-REASON	89.4	-	-
22	Qwen: Qwen3.8 27B	89.2	<$0.01	-
23	Qwen3.8-27B	89.2	-	-
24	Ornith-1.5-35B-A3B	89.2	-	-
25	Darwin-36B-Opus	88.4	-	-
26	Qwen3.5-397B-A17B	88.4	-	-
27	Qwen: Qwen3.5 397B A17B	88.4	<$0.01	-
28	Darwin-60B-DUO	88.4	-	-
29	Ring-2.6-1T	88.3	-	-
30	DeepSeek-V4-Flash	88.1	-	-
31	NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4	87.9	-	-
32	GLM-4.7-FP8	87.9	-	-
33	Qwen: Qwen3.6 27B	87.8	<$0.01	-
34	Qwen3.6-27B	87.8	-	-
35	MoonshotAI: Kimi K2.5	87.6	<$0.01	-
36	Kimi-K2.5	87.6	-	-
37	MoonshotAI: Kimi K2.5	87.4	<$0.01	-
38	Kimi-K2.5	87.4	-	-
39	Hy3-preview	87.2	-	-
40	Inkling	87.2	-	-
41	Thinking Machines: Inkling	87.2	<$0.01	-
42	Tencent: Hy3 preview	87.2	<$0.01	-
43	NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16	87.0	-	-
44	Darwin-27B-Opus	86.9	-	-
45	Qwen: Qwen3.5-122B-A10B	86.6	<$0.01	-
46	Qwen3.5-122B-A10B	86.6	-	-
47	Ornith-1.5-9B	86.4	-	-
48	Ourbox-35B-JGOS	86.4	-	-
49	Solar-Open2-250B	86.3	-	-
50	GLM-5.1	86.2	-	-
51	GLM-5.1-MXFP4-Mixed-CT-AutoRound	86.2	-	-
52	GLM-5	86.0	-	-
53	Qwen3.6-35B-A3B	86.0	-	-
54	Qwen: Qwen3.6 35B A3B	86.0	<$0.01	-
55	Darwin-31B-Opus	85.9	-	-
56	GLM-4.7	85.7	-	-
57	A.X-K2	85.6	-	-
58	Qwen3.5-27B	85.5	-	-
59	Qwen: Qwen3.5-27B	85.5	<$0.01	-
60	GLM-5-FP8	85.4	-	-
61	MiniMax-M2.5	85.2	-	-
62	Darwin-4B-David	85.0	-	-
63	Qwen: Qwen3.5-122B-A10B	84.8	<$0.01	-
64	Qwen3.5-122B-A10B	84.8	-	-
65	Kimi-K2-Thinking	84.5	-	-
66	MoonshotAI: Kimi K2 Thinking	84.5	<$0.01	-
67	Darwin-9B-NEG	84.3	-	-
68	JGOS-31B-Citizen	84.3	-	-
69	openPangu-2.0-Flash	83.7	-	-
70	Muse-Glimmer-30B	83.5	-	-
71	Step-3.5-Flash	83.5	-	-
72	GLM-5.1-FP8	83.3	-	-
73	MiMo-V2-Flash	83.3	-	-
74	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	82.7	-	-
75	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	82.7	-	-
76	Qwen: Qwen3.5-27B	81.8	<$0.01	-
77	Qwen3.5-27B	81.8	-	-
78	GLM-4.6-FP8	81.3	-	-
79	gpt-oss-120b	80.9	<$0.01	-
80	OpenAI: gpt-oss-120b	80.9	<$0.01	-
81	MiniMax-M2.1	80.8	-	-
82	OpenAI: gpt-oss-120b	80.8	<$0.01	-
83	gpt-oss-120b	80.8	<$0.01	-
84	EXAONE-4.5-33B	80.5	-	-
85	LongCat-Flash-Thinking-2601	80.5	-	-
86	OpenAI: gpt-oss-120b	80.1	<$0.01	-
87	gpt-oss-120b	80.1	<$0.01	-
88	openPangu-2.0-Flash	79.8	-	-
89	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	79.2	-	-
90	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	79.2	-	-
91	K-EXAONE-236B-A23B	79.1	-	-
92	DeepSeek-R1-0528	78.8	-	-
93	NVIDIA-Nemotron-3-Super-120B-A12B-FP8	77.3	-	-
94	Nemotron-Cascade-2-30B-A3B	76.1	-	-
95	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4	75.6	-	-
96	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16	75.4	-	-
97	GLM-4.7-Flash	75.2	-	-
98	JoyAI-LLM-Flash	74.4	-	-
99	gpt-oss-20b	74.2	<$0.01	-
100	OpenAI: gpt-oss-20b	74.2	<$0.01	-
101	A.X-K1	74.0	-	-
102	OpenAI: gpt-oss-120b	73.5	<$0.01	-
103	gpt-oss-120b	73.5	<$0.01	-
104	OpenAI: gpt-oss-120b	73.1	<$0.01	-
105	gpt-oss-120b	73.1	<$0.01	-
106	Qwen: Qwen3 30B A3B Thinking 2507	72.2	<$0.01	-
107	Qwen3-30B-A3B-Thinking-2507	72.2	-	-
108	GLM-4.5-Air	71.7	-	-
109	OpenAI: gpt-oss-20b	71.5	<$0.01	-
110	DeepSeek-R1	71.5	-	-
111	gpt-oss-20b	71.5	<$0.01	-
112	Mistral-Small-4-119B-2603	71.2	-	-
113	OpenAI: gpt-oss-120b	68.1	<$0.01	-
114	gpt-oss-120b	68.1	<$0.01	-
115	OpenAI: gpt-oss-20b	67.1	<$0.01	-
116	gpt-oss-20b	67.1	<$0.01	-
117	OpenAI: gpt-oss-120b	67.1	<$0.01	-
118	gpt-oss-120b	67.1	<$0.01	-
119	MiMo-V2.5-Pro	66.7	-	-
120	gpt-oss-20b	66.0	<$0.01	-
121	OpenAI: gpt-oss-20b	66.0	<$0.01	-
122	Qwen3-4B-Thinking-2507	65.8	-	-
123	Qwen3-4B-Instruct-2507	62.0	-	-
124	OpenAI: gpt-oss-20b	58.6	<$0.01	-
125	gpt-oss-20b	58.6	<$0.01	-
126	DeepSeek-V3	58.2	-	-
127	OpenAI: gpt-oss-20b	58.0	<$0.01	-
128	gpt-oss-20b	58.0	<$0.01	-
129	Mellum2-12B-A2.5B-Thinking	57.6	-	-
130	gpt-oss-20b	56.8	<$0.01	-
131	OpenAI: gpt-oss-20b	56.8	<$0.01	-
132	c4ai-command-a-03-2025	50.5	-	-
133	Llama-3.2-90B-Vision-Instruct	46.1	-	-
134	granite-4.1-30b	45.8	-	-
135	granite-4.1-8b	42.0	-	-
136	Mellum2-12B-A2.5B-Instruct	40.9	-	-
137	Mellum2-12B-A2.5B-Thinking-SFT	39.9	-	-
138	Mellum2-12B-A2.5B-Instruct-SFT	38.9	-	-
139	c4ai-command-r-plus-08-2024	34.3	-	-
140	granite-4.1-3b	31.7	-	-
141	Mellum2-12B-A2.5B-Base	31.3	-	-
142	Mellum2-12B-A2.5B-Base-Pretrain	31.3	-	-
143	c4ai-command-r7b-12-2024	26.8	-	-
144	c4ai-command-r-08-2024	26.8	-	-
145	Llama-3.2-1B-Instruct	18.7	-	-
146	Meta: Llama 3.2 1B Instruct	18.7	<$0.01	-

benchmark

MMLU-Pro

Ranked by published score, highest first.

#	Model	Score	Price / 1k tokens	Response time
1	Darwin-180B-RSI	88.1	-	-
2	Intern-S2-Preview	88.0	-	-
3	MiniMax-M2.1	88.0	-	-
4	Qwen: Qwen3.5 397B A17B	87.8	<$0.01	-
5	Qwen3.5-397B-A17B	87.8	-	-
6	DeepSeek-V4-Pro	87.5	-	-
7	Kimi-K2.5	87.1	-	-
8	MoonshotAI: Kimi K2.5	87.1	<$0.01	-
9	NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4	86.8	-	-
10	NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16	86.8	-	-
11	Qwen: Qwen3.5-122B-A10B	86.7	<$0.01	-
12	Qwen3.5-122B-A10B	86.7	-	-
13	DeepSeek-V4-Flash	86.4	-	-
14	Qwen: Qwen3.6 27B	86.2	<$0.01	-
15	Qwen3.6-27B	86.2	-	-
16	Solar-Open2-250B	86.2	-	-
17	Qwen: Qwen3.5-27B	86.1	<$0.01	-
18	Qwen3.5-27B	86.1	-	-
19	GLM-5	86.0	-	-
20	Qwen3.6-35B-A3B	85.2	-	-
21	Qwen: Qwen3.6 35B A3B	85.2	<$0.01	-
22	DeepSeek-R1-0528	85.0	-	-
23	GLM-4.5	84.6	-	-
24	Step-3.5-Flash	84.4	-	-
25	DeepSeek-R1	84.0	-	-
26	K-EXAONE-236B-A23B	83.8	-	-
27	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	83.7	-	-
28	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	83.7	-	-
29	Intern-S1	83.5	-	-
30	EXAONE-4.5-33B	83.3	-	-
31	Qwen3-235B-A22B-Instruct-2507	83.0	-	-
32	LongCat-Flash-Chat	82.7	-	-
33	Seed-OSS-36B-Instruct	82.7	-	-
34	MiniMax-M2	82.0	-	-
35	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16	81.9	-	-
36	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4	81.6	-	-
37	A.X-K1	81.5	-	-
38	GLM-4.5-Air	81.4	-	-
39	DeepSeek-V3-0324	81.2	-	-
40	MiniMax-M1-40k	81.1	-	-
41	JoyAI-LLM-Flash	81.0	-	-
42	Kimi-K2-Instruct	81.0	-	-
43	Qwen: Qwen3 30B A3B Thinking 2507	80.9	<$0.01	-
44	Qwen3-30B-A3B-Thinking-2507	80.9	-	-
45	OpenAI: gpt-oss-120b	80.8	<$0.01	-
46	gpt-oss-120b	80.8	<$0.01	-
47	MiniMax-M2.5	80.1	-	-
48	ERNIE-4.5-300B-A47B-PT	78.4	-	-
49	LongCat-Flash-Lite	78.3	-	-
50	MiniMax-Text-01	75.7	-	-
51	OpenAI: gpt-oss-20b	73.6	<$0.01	-
52	gpt-oss-20b	73.6	<$0.01	-
53	Qwen2.5-72B	71.6	-	-
54	Microsoft: Phi 4	70.4	<$0.01	-
55	phi-4	70.4	<$0.01	-
56	Qwen3-4B-Instruct-2507	69.6	-	-
57	ERNIE-4.5-300B-A47B-Base-PT	69.5	-	-
58	Qwen2.5-32B	69.2	-	-
59	MiMo-V2.5-Pro	68.5	-	-
60	Qwen: Qwen3 235B A22B	68.2	<$0.01	-
61	Qwen3-235B-A22B	68.2	-	-
62	Mistral-Large-Instruct-2411	67.9	-	-
63	Hunyuan-A13B-Instruct	67.3	-	-
64	Tencent: Hunyuan A13B Instruct	67.3	<$0.01	-
65	Mistral-Large-Instruct-2407	65.9	-	-
66	DeepSeek-V2.5	65.8	-	-
67	NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16	65.1	-	-
68	Seed-OSS-36B-Base	65.1	-	-
69	DeepSeek-V3	64.4	-	-
70	granite-4.1-30b	64.1	-	-
71	Qwen2.5-14B	63.7	-	-
72	Qwen3-30B-A3B-Base	61.7	-	-
73	Llama-3.1-405B	61.6	-	-
74	Nemotron-H-56B-Base-8K	60.5	-	-
75	Seed-OSS-36B-Base-woSyn	60.4	-	-
76	Tencent-Hunyuan-Large	60.2	-	-
77	Mellum2-12B-A2.5B-Base	59.3	-	-
78	Mellum2-12B-A2.5B-Base-Pretrain	59.3	-	-
79	EXAONE-3.5-32B-Instruct	58.9	-	-
80	MiMo-7B-RL	58.6	-	-
81	internlm3-8b-instruct	57.6	-	-
82	ERNIE-4.5-21B-A3B-Base-PT	56.7	-	-
83	granite-4.1-8b	56.0	-	-
84	Phi-3-medium-4k-instruct	55.7	-	-
85	DeepSeek-V2-Chat	54.8	-	-
86	Mistral-Small-24B-Base-2501	54.4	-	-
87	Phi-4-mini-instruct	52.8	-	-
88	Meta-Llama-3-70B	52.8	-	-
89	Llama-3.1-70B	52.5	-	-
90	Yi-1.5-34B-Chat	52.3	-	-
91	Phi-3-medium-128k-instruct	51.9	-	-
92	MAmmoTH2-8x7B-Plus	50.4	-	-
93	Qwen1.5-110B	49.9	-	-
94	granite-4.1-3b	49.8	-	-
95	AI21-Jamba-Large-1.5	49.5	-	-
96	Mistral-Small-Instruct-2409	48.4	-	-
97	glm-4-9b	47.9	-	-
98	Phi-3.5-mini-instruct	47.9	-	-
99	EXAONE-3.5-7.8B-Instruct	46.2	-	-
100	Yi-1.5-9B-Chat	46.0	-	-
101	Phi-3-mini-4k-instruct	45.7	-	-
102	aya-expanse-32b	45.4	-	-
103	gemma-2-9b	45.1	-	-
104	Qwen2.5-7B	45.0	-	-
105	Phi-3-mini-128k-instruct	43.9	-	-
106	Qwen2.5-3B	43.7	-	-
107	MAmmoTH2-8B-Plus	43.4	-	-
108	Yi-34B	43.0	-	-
109	Mathstral-7B-v0.1	42.0	-	-
110	MiMo-7B-Base	41.9	-	-
111	DeepSeek-Coder-V2-Lite-Instruct	41.6	-	-
112	Mixtral-8x7B-v0.1	41.0	-	-
113	Meta-Llama-3-8B-Instruct	41.0	-	-
114	MAmmoTH2-7B-Plus	40.9	-	-
115	Qwen2-7B	40.7	-	-
116	Mistral-Nemo-Base-2407	39.8	-	-
117	EXAONE-3.5-2.4B-Instruct	39.1	-	-
118	Yi-1.5-6B-Chat	38.2	-	-
119	Qwen1.5-14B-Chat	38.0	-	-
120	Ministral-8B-Instruct-2410	37.9	-	-
121	c4ai-command-r-v01	37.9	-	-
122	internlm2-math-plus-20b	37.1	-	-
123	LLaDA-8B-Instruct	37.0	-	-
124	Llama-3-Smaug-8B	36.9	-	-
125	Llama-3.1-8B	36.6	-	-
126	Meta-Llama-3-8B	35.4	-	-
127	deepseek-math-7b-instruct	35.3	-	-
128	DeepSeek-Coder-V2-Lite-Base	34.4	-	-
129	aya-expanse-8b	33.7	-	-
130	internlm2-math-plus-7b	33.5	-	-
131	granite-3.1-8b-base	33.1	-	-
132	Qwen2.5-1.5B	32.1	-	-
133	granite-3.0-8b-base	31.0	-	-
134	Mistral-7B-Instruct-v0.2	30.8	-	-
135	Mistral-7B-v0.2	30.4	-	-
136	Qwen1.5-7B-Chat	29.1	-	-
137	Yi-6B-Chat	28.8	-	-
138	Yi-6B	26.5	-	-
139	granite-3.1-2b-base	23.9	-	-
140	llemma_7b	23.4	-	-
141	Qwen2-1.5B-Instruct	22.6	-	-
142	Qwen2-1.5B	22.6	-	-
143	Llama-3.2-3B	22.2	-	-
144	granite-3.0-2b-base	21.7	-	-
145	granite-3.1-3b-a800m-base	20.4	-	-
146	SmolLM2-1.7B	18.3	-	-
147	Qwen2-0.5B	15.0	-	-
148	Qwen2.5-0.5B	14.9	-	-
149	granite-3.1-1b-a400m-base	12.3	-	-
150	Llama-3.2-1B	11.9	-	-
151	SmolLM2-360M	11.4	-	-
152	SmolLM2-135M	10.8	-	-
153	Qwen: Qwen2.5 VL 72B Instruct	0.6	<$0.01	-
154	Qwen2.5-VL-72B-Instruct	0.6	-	-

benchmark

SWE-bench Verified

Ranked by published score, highest first.

#	Model	Score	Price / 1k tokens	Response time
1	Ornith-1.5-397B	86.0	-	-
2	Macaron-V1-Coding-Venti	85.6	-	-
3	Macaron-V1-Venti	85.6	-	-
4	Ornith-1.0-397B	82.4	-	-
5	DeepSeek-V4-Pro	80.6	-	-
6	MiniMax-M3	80.5	-	-
7	MoonshotAI: Kimi K2.6	80.2	<$0.01	-
8	Thinking Machines: Inkling Small	80.2	<$0.01	-
9	Kimi-K2.6	80.2	-	-
10	Inkling-Small	80.2	-	-
11	DeepSeek-V4-Flash	79.0	-	-
12	Ornith-1.5-35B-A3B	79.0	-	-
13	MiMo-V2.5-Pro	78.9	-	-
14	dots3-note-prev	78.4	-	-
15	Hy3	78.0	-	-
16	Tencent: Hy3	78.0	<$0.01	-
17	GLM-5	77.8	-	-
18	Mistral-Medium-3.5-128B	77.6	-	-
19	Inkling	77.6	-	-
20	Thinking Machines: Inkling	77.6	<$0.01	-
21	Qwen: Qwen3.6 27B	77.2	<$0.01	-
22	Qwen3.6-27B	77.2	-	-
23	Qwen: Qwen3.5 397B A17B	76.4	<$0.01	-
24	Qwen3.5-397B-A17B	76.4	-	-
25	Muse-Glimmer-30B	76.0	-	-
26	MiniMax-M2.5	75.8	-	-
27	Ornith-1.0-35B	75.6	-	-
28	Macaron-V1-Tall	75.4	-	-
29	IndustrialCoder	74.8	-	-
30	Laguna-M.1	74.6	-	-
31	Step-3.5-Flash	74.4	-	-
32	Tencent: Hy3 preview	74.4	<$0.01	-
33	Hy3-preview	74.4	-	-
34	Ring-2.6-1T	74.0	-	-
35	MiniMax-M2.1	74.0	-	-
36	GLM-4.7	73.8	-	-
37	Qwen: Qwen3.6 35B A3B	73.4	<$0.01	-
38	Qwen3.6-35B-A3B	73.4	-	-
39	GLM-5	72.8	-	-
40	Qwen: Qwen3.5-27B	72.4	<$0.01	-
41	Qwen3.5-27B	72.4	-	-
42	Ling-2.6-1T	72.2	-	-
43	Qwen3.5-122B-A10B	72.0	-	-
44	Qwen: Qwen3.5-122B-A10B	72.0	<$0.01	-
45	NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16	71.9	-	-
46	MoonshotAI: Kimi K2 Thinking	71.3	<$0.01	-
47	Kimi-K2-Thinking	71.3	-	-
48	Laguna-XS-2.1	70.9	-	-
49	Poolside: Laguna XS 2.1	70.9	<$0.01	-
50	MoonshotAI: Kimi K2.5	70.8	<$0.01	-
51	Kimi-K2.5	70.8	-	-
52	Qwen3-Coder-Next	70.6	-	-
53	Qwen: Qwen3 Coder Next	70.6	<$0.01	-
54	Ornith-1.5-9B	70.6	-	-
55	Solar-Open2-250B	70.4	-	-
56	Laguna-XS.2	69.9	-	-
57	NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4	69.7	-	-
58	Ornith-1.0-9B	69.4	-	-
59	MiniMax-M2	69.4	-	-
60	Laguna-XS.2	68.2	-	-
61	North-Mini-Code-1.0	67.6	-	-
62	Intern-S2-Preview	64.0	-	-
63	openPangu-2.0-Flash	63.1	-	-
64	gpt-oss-120b	62.4	<$0.01	-
65	OpenAI: gpt-oss-120b	62.4	<$0.01	-
66	Ling-2.6-flash	61.2	-	-
67	gpt-oss-20b	60.7	<$0.01	-
68	OpenAI: gpt-oss-20b	60.7	<$0.01	-
69	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	60.5	-	-
70	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	60.5	-	-
71	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	59.2	-	-
72	GLM-4.7-Flash	59.2	-	-
73	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	59.2	-	-
74	openPangu-2.0-Flash	57.6	-	-
75	cwm	53.9	-	-
76	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	53.7	-	-
77	NVIDIA-Nemotron-3-Super-120B-A12B-BF16	53.7	-	-
78	gpt-oss-20b	53.2	<$0.01	-
79	OpenAI: gpt-oss-20b	53.2	<$0.01	-
80	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4	52.8	-	-
81	OpenAI: gpt-oss-120b	52.6	<$0.01	-
82	gpt-oss-120b	52.6	<$0.01	-
83	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16	51.6	-	-
84	OpenAI: gpt-oss-120b	47.9	<$0.01	-
85	gpt-oss-120b	47.9	<$0.01	-
86	OpenAI: gpt-oss-20b	37.4	<$0.01	-
87	gpt-oss-20b	37.4	<$0.01	-

For crawlers

Read the ranking as JSON.

One request, the same rows this page shows.

GET /api/v1/models/best