Best LLM APIs by intelligence per dollar

A second take on value, from a different benchmark. This ranks 111 models by the Epoch Capabilities Index — Epoch AI's capability score, combined from dozens of benchmarks (higher is better, the scale has no zero; CC BY 4.0) — divided by each model's cheapest blended price per 1M tokens. Where the Elo value board measures human preference, this measures aggregated reasoning and knowledge. Each model links to its cheapest provider; all figures computed from the index.

Ranked by intelligence per dollar

#	Model	Index	Cheapest /1M	Verified	Index / $
1	qwen3.7-flash · Alibaba	144	$0.16	today	900
2	glm-5.3-flash · StreamLake	153	$0.189	today	811.7
3	deepseek-v4-flash · Relace	155	$0.21	today	738.1
4	deepseek-v4-flash-0423 · StreamLake	146	$0.21	today	695.2
5	qwen3.5-9b · SiliconFlow	139	$0.25	today	556
6	gemma-4-26b-a4b · Reka	142	$0.26	23d ago	546.2
7	gemma-4-31b · Reka	143	$0.38	24d ago	376.3
8	deepseek-v4.1-flash · Io Net	156	$0.468	21d ago	333.3
9	qwen3.5-flash · Alibaba	144	$0.5	today	288
10	deepseek-v3.2 · GMICloud	146	$0.518	today	281.6
11	claude-haiku-5-5 · Anthropic	159	$0.6	16d ago	265
12	gpt-6-luna · OpenAI	157	$0.6	25d ago	261.7
13	qwen3.6-35b-a3b · AkashML	144	$1	today	144
14	minimax-m2.7 · GMICloud	146	$1.05	today	139
15	qwen3.5-35b-a3b · DeepInfra	142	$1.14	today	124.6
16	minimax-m3 · CoreWeave	147	$1.19	today	123.5
17	minimax-m2.5 · Venice	147	$1.22	today	120.5
18	qwen3.8-27b · Chutes	152	$1.289	today	117.9
19	gpt-5.6-luna · OpenAI	156	$1.4	today	111.4
20	inkling-small · DeepInfra	150	$1.65	today	90.9
21	qwen3.6-flash · Alibaba	143	$1.75	today	81.7
22	qwen3.7-plus · Alibaba	147	$2	today	73.5
23	glm-4.7 · DeepInfra	144	$2.15	today	67
24	glm-5.3 · Inceptron	156	$2.36	today	66.1
25	qwen3.6-27b · Chutes	146	$2.25	today	64.9
26	glm-5.2 · DeepInfra	152	$2.362	today	64.3
27	kimi-k2.6 · Chutes	151	$2.4	today	62.9
28	deepseek-v4 · Relace	156	$2.5	today	62.4
29	deepseek-v4-pro-0423 · Relace	149	$2.4	today	62.1
30	glm-5 · GMICloud	146	$2.52	today	57.9
31	kimi-k2.5 · DeepInfra	148	$2.7	46d ago ⚠	54.8
32	nemotron-3-ultra-550b-a55b · DeepInfra	146	$2.7	today	54.1
33	qwen3.5-plus · Alibaba	146	$2.8	today	52.1
34	gemini-3.5-flash-lite · Google	145	$2.8	48d ago ⚠	51.8
35	kimi-k2-thinking · Google	146	$3.1	today	47.1
36	gemini-3-flash-preview · Google	152	$3.5	6d ago	43.4
37	qwen3.5-397b-a17b · DeepInfra	147	$3.45	today	42.6
38	qwen3.6-plus · Alibaba	147	$3.5	today	42
39	kimi-k2.7-code · StreamLake	150	$3.712	today	40.4
40	glm-5.1 · StreamLake	150	$4.002	today	37.5
41	gemini-3.7-flash · Google	158	$4.5	48d ago ⚠	35.1
42	gemini-3.8-flash · Google	157	$4.5	6d ago	34.9
43	gemini-3.6-flash · Google	155	$4.5	50d ago ⚠	34.4
44	qwen3.7-max · Novita AI	154	$5	today	30.8
45	inkling · DeepInfra	148	$5	1d ago	29.6
46	muse-spark-1.3 · Meta	157	$5.5	today	28.5
47	muse-spark-open · Together AI	155	$5.5	today	28.2
48	mistral-large-4 · Mistral AI	150	$5.54	today	27.1
49	grok-4.6 · xAI	157	$8	6d ago	19.6
50	qwen3.8-max · Novita AI	157	$8	today	19.6
51	grok-4.7 · xAI	154	$8	29d ago	19.2
52	grok-4.5 · xAI	154	$8	48d ago ⚠	19.2
53	qwen3.6-max-preview · Alibaba	149	$9.1	today	16.4
54	mistral-medium-3-5 · Mistral AI	141	$9	today	15.7
55	gpt-6.1-sol · OpenAI	168	$12	24d ago	14
56	kimi-k3 · Chutes	158	$11.35	today	13.9
57	claude-sonnet-5-5 · Anthropic	166	$12	6d ago	13.8
58	gemini-4-argon · Google	166	$12	6d ago	13.8
59	gpt-6-sol · OpenAI	164	$12	25d ago	13.7
60	claude-sonnet-5 · Anthropic	157	$12	today	13.1
61	gpt-5.6-terra · OpenAI	160	$14	today	11.4
62	gemini-3.1-pro-preview · Google	155	$14	6d ago	11.1
63	gpt-5.3-codex · OpenAI	157	$15.75	6d ago	10
64	claude-opus-5-5 · Anthropic	169	$24	29d ago	7
65	gpt-5.6-sol · OpenAI	163	$24	today	6.8
66	muse-wonder · Meta	162	$24	today	6.8
67	claude-opus-5 · Anthropic	164	$30	today	5.5
68	claude-fable-5-1 · Anthropic	166	$60	48d ago ⚠	2.8
69	gpt-6-astra · OpenAI	169	$60	6d ago	2.8
70	claude-fable-5 · Anthropic	163	$60	today	2.7

Value = index score ÷ (input + output price per 1M tokens). A model's score is identical across providers, so the cheapest host gives the best value — that's the price shown. Models released within the last 12 months; older ones below. Blended is a coarse ranking aid, not a workload estimate; size your own on each model's page.

Older models — released more than 12 months ago (41)

Still priced daily and often still the pragmatic choice, but no longer the current generation.

#	Model	Index	Cheapest /1M	Age	Index / $
71	mistral-nemo · Io Net	119	$0.038	27 mo	3,115.2
72	llama-3.1-8b · DeepInfra	116	$0.06	27 mo	1,933.3
73	gemma-3-4b · Parasail	116	$0.09	19 mo	1,288.9
74	gpt-oss-20b · AkashML	138	$0.12	14 mo	1,150
75	gemma-3-12b · Novita AI	124	$0.15	19 mo	826.7
76	gpt-oss-120b · Venice	140	$0.18	14 mo	777.8
77	phi-4 · DeepInfra	130	$0.21	22 mo	619
78	qwen3-30b-a3b · StreamLake	136	$0.241	17 mo	564.1
79	gemma-3-27b · DeepInfra	130	$0.24	19 mo	541.7
80	deepseek-v3.2-exp · AtlasCloud	145	$0.336	12 mo	431.5
81	qwen3-14b · NextBit	138	$0.32	17 mo	431.2
82	qwen-2.5-7b · Phala	119	$0.3	24 mo	396.7
83	qwen3-32b · DeepInfra	138	$0.36	17 mo	383.3
84	llama-4-scout · DeepInfra	130	$0.4	18 mo	325
85	qwen3-235b-a22b · GMICloud	139	$0.438	15 mo	317.7
86	llama-3.3-70b · DeepInfra	127	$0.42	22 mo	302.4
87	qwen-2.5-72b · DeepInfra	129	$0.76	25 mo	169.7
88	gpt-4o-mini · OpenAI	126	$0.75	27 mo	168
89	llama-3.1-70b · DeepInfra	126	$0.8	27 mo	157.5
90	qwen3-8b · Alibaba	136	$0.88	17 mo	154.5
91	mistral-small-3.1-24b · Cloudflare	128	$0.906	19 mo	141.3
92	llama-4-maverick · DeepInfra	132	$1	18 mo	132
93	deepseek-chat · DeepInfra	146	$1.21	21 mo	120.7
94	deepseek-chat-v3.1 · DeepInfra	140	$1.2	14 mo	116.7
95	deepseek-v3 · DeepInfra	132	$1.21	21 mo	109.1
96	gemma-2-27b · NextBit	122	$1.3	27 mo	93.8
97	qwen-2.5-coder-32b · Cloudflare	119	$1.66	23 mo	71.7
98	gpt-4.1-mini · OpenAI	135	$2	18 mo	67.5
99	qwen3-235b-a22b-thinking · Alibaba	144	$2.53	14 mo	56.9
100	mistral-medium-3 · Mistral AI	134	$2.4	17 mo	55.8
101	qwen3-30b-a3b-thinking · Alibaba	140	$2.6	14 mo	53.8
102	deepseek-r1 · DeepInfra	139	$2.65	16 mo	52.5
103	gemini-2.5-flash · Google	141	$2.8	16 mo	50.4
104	claude-haiku-4-5 · Anthropic	142	$6	12 mo	23.7
105	qwen3-max · Alibaba	142	$7.2	12 mo	19.7
106	mixtral-8x22b · Mistral AI	122	$8	30 mo	15.2
107	gpt-4.1 · OpenAI	137	$10	18 mo	13.7
108	gemini-2.5-pro · Google	145	$11.25	16 mo	12.9
109	command-r-plus · Cohere	119	$12.5	25 mo	9.5
110	claude-sonnet-4-5 · Anthropic	147	$18	12 mo	8.2
111	claude-opus-4-1 · Anthropic	144	$90	14 mo	1.6

For contrast: the top of the intelligence index

The highest-scoring models by raw index — rarely the best value, which is the point.

Index rank	Model	Index	Cheapest /1M	Index / $
1	claude-opus-5-5	169	$24	7
2	gpt-6-astra	169	$60	2.8
3	gpt-6.1-sol	168	$12	14
4	claude-fable-5-1	166	$60	2.8
5	claude-sonnet-5-5	166	$12	13.8
6	gemini-4-argon	166	$12	13.8
7	claude-opus-5	164	$30	5.5
8	gpt-6-sol	164	$12	13.7
9	claude-fable-5	163	$60	2.7
10	gpt-5.6-sol	163	$24	6.8

Epoch Capabilities Index · by Epoch AI, CC BY 4.0 · read 2026-10-23 · source · full methodology. Two benchmarks disagree by design — cross-check the Elo value ranking.

FAQ

Which LLM API gives the most intelligence per dollar?

As of 2026-10-23, qwen3.7-flash leads: 900 index points per dollar — a 144 Epoch Capabilities Index score at $0.16 per 1M tokens (blended input + output) via Alibaba.

How is this different from the Elo value ranking?

This ranks by the Epoch Capabilities Index from Epoch AI — one score per model, combined from dozens of benchmarks and published under CC BY 4.0 — instead of LMArena Elo, which measures human preference. Two independent measures cross-check each other; they disagree by design, so a model strong on one may rank differently on the other.

Is the smartest model the best value?

Rarely. The top model by raw index is claude-opus-5-5 at 169, but at $24 per 1M its value is 7 points per dollar — below the value leader qwen3.7-flash. Peak intelligence usually carries a price premium.