API host · catalogue checked 27 Oct 2026

DeepInfra: models, prices and uptime

DeepInfra serves 26 of the 78 models we benchmark through OpenRouter, 3 of them on no other host. Prices are OpenRouter's list prices per million tokens. We did not measure speed per host.

Models served26of 78 we test

Cheapest host for7of 23 shared models

Median premium+24%over the cheapest host

Uptime99.4%last 30 min, average

Most accurate models on DeepInfra

Verification score of each model this host serves

1DeepSeek V4.1 Flash 1017DeepSeek73.8%
2DeepSeek V4 Flash Vision ExpDeepSeek71.2%
3DeepSeek V4 Pro 0813DeepSeek71.2%
4MiMo-V2.6-ProXiaomi68.8%
5GLM 5.3 FlashZ.ai67.5%
6Inkling SmallThinking Machines67.5%
7GLM 5.3Z.ai67.5%
8Step 5 PreviewStepFun63.8%
9Qwen3.8 27BQwen63.8%
10MiMo-V2.6-FlashXiaomi62.5%
11Nemotron 3 UltraNVIDIA60.0%
12gpt-oss-120bOpenAI58.8%
13MiniMax M3MiniMax57.5%
14gpt-oss-20bOpenAI51.2%
15Nemotron 3.5 LightningNVIDIA43.8%
16Llama 3.3 70B InstructMeta42.5%
17Llama 4 ScoutMeta41.2%
18Ling 3.1 FlashinclusionAI41.2%
19Phi 4Microsoft41.2%
20Nemotron 3.5 Super VLNVIDIA40.0%
21Qwen3.6 35B A3BQwen37.5%
22Granite 4.2 8BIBM36.2%
23Qwen3.5 397B A17BQwen32.5%
24Gemma 4 31BGoogle30.0%
25Gemma 4 26B A4BGoogle15.0%

Models on DeepInfra

Premium: this host's blended price (3 input : 1 output) against the cheapest host for the same model

Model	Score	Input / 1M	Output / 1M	vs cheapest host	Hosts	Context	Max output	Quantization	Uptime
DeepSeek V4.1 Flash 1017	73.8%	$0.0784	$0.2352	+4%	30	1M	131k	fp8	86.7%
DeepSeek V4 Flash Vision Exp	71.2%	$0.2156	$0.6468	Cheapest	3	1M	262k	fp8	100%
DeepSeek V4 Pro 0813	71.2%	$1.3	$2.6	+674%	23	1M	16k	fp8	100%
MiMo-V2.6-Pro	68.8%	$0.387	$0.783	+5%	5	1M	944k	fp8	100%
GLM 5.3 Flash	67.5%	$0.15	$0.5	+406%	34	1M	944k	fp4	100%
Inkling Small	67.5%	$0.45	$1.2	Only host	1	524k	262k	fp8	100%
GLM 5.3	67.5%	$0.5625	$2.5	+160%	37	1M	131k	fp4	100%
Step 5 Preview	63.8%	$1	$2.7	Cheapest	4	1M	66k	fp8	100%
Qwen3.8 27B	63.8%	$0.15	$1.875	+242%	21	262k	236k	bf16	100%
MiMo-V2.6-Flash	62.5%	$0.14	$0.28	+21%	10	1M	944k	fp8	100%
Nemotron 3 Ultra	60.0%	$0.4	$1.76	Cheapest	3	262k	16k	fp4	—
gpt-oss-120b	58.8%	$0.037	$0.17	+17%	23	131k	118k	bf16	100%
MiniMax M3	57.5%	$0.224	$0.88	Cheapest	12	524k	131k	fp8	—
gpt-oss-20b	51.2%	$0.03	$0.14	+60%	10	131k	118k	bf16	100%
Nemotron 3.5 Lightning	43.8%	$0.06	$0.16	+74%	4	262k	33k	bf16	100%
Llama 3.3 70B Instruct	42.5%	$0.1	$0.32	Cheapest	9	131k	16k	fp8	100%
Llama 4 Scout	41.2%	$0.08	$0.24	Cheapest	2	328k	295k	fp8	—
Ling 3.1 Flash	41.2%	$0.075	$0.225	Cheapest	2	262k	236k	fp4	100%
Phi 4	41.2%	$0.056	$0.112	Only host	1	16k	15k	bf16	—
Nemotron 3.5 Super VL	40.0%	$0.15	$0.5	+79%	9	262k	236k	fp4	100%
Qwen3.6 35B A3B	37.5%	$0.1	$0.95	+47%	10	262k	66k	fp8	100%
Granite 4.2 8B	36.2%	$0.06	$0.25	+30%	2	131k	118k	bf16	100%
Qwen3.5 397B A17B	32.5%	$0.45	$3	+24%	10	262k	82k	fp8	100%
Gemma 4 31B	30.0%	$0.16	$0.32	+25%	13	262k	236k	fp8	—
Gemma 4 26B A4B	15.0%	$0.07	$0.34	+59%	15	262k	16k	fp8	100%
Ling 3.0 Flash VL	—	$0.06	$0.18	Only host	1	131k	33k	fp16	100%

All API hosts →

DeepInfra: frequently asked questions

Which AI models does DeepInfra serve?: Of the models we benchmark, DeepInfra serves 26 through OpenRouter: DeepSeek V4.1 Flash 1017, DeepSeek V4 Flash Vision Exp, DeepSeek V4 Pro 0813, MiMo-V2.6-Pro, GLM 5.3 Flash, Inkling Small, GLM 5.3, Step 5 Preview, Qwen3.8 27B, MiMo-V2.6-Flash, Nemotron 3 Ultra, gpt-oss-120b, MiniMax M3, gpt-oss-20b, Nemotron 3.5 Lightning, Llama 3.3 70B Instruct, Llama 4 Scout, Ling 3.1 Flash, Phi 4, Nemotron 3.5 Super VL, Qwen3.6 35B A3B, Granite 4.2 8B, Qwen3.5 397B A17B, Gemma 4 31B, Gemma 4 26B A4B, Ling 3.0 Flash VL.
What is the most accurate model on DeepInfra?: DeepSeek V4.1 Flash 1017 by DeepSeek, with a verification score of 73.8% (59 of 80 questions correct) in our tests.
Is DeepInfra the cheapest host?: DeepInfra has the lowest list price for 7 of the 23 models it serves that other hosts also serve (DeepSeek V4 Flash Vision Exp, Step 5 Preview, Nemotron 3 Ultra, MiniMax M3, Llama 3.3 70B Instruct, Llama 4 Scout, Ling 3.1 Flash). Its median markup over the cheapest host is +24%.
How reliable is DeepInfra?: Averaged over the models it serves, DeepInfra reported 99.4% uptime in the 30 minutes before OpenRouter's catalogue was read for our run.

Model

Score

Input / 1M

Output / 1M

vs cheapest host

Hosts

Context

Max output

Quantization

Uptime

DeepSeek V4.1 Flash 1017

73.8%

$0.0784

$0.2352

+4%

131k

fp8

86.7%

DeepSeek V4 Flash Vision Exp

71.2%

$0.2156

$0.6468

Cheapest

262k

fp8

100%

DeepSeek V4 Pro 0813

71.2%

$1.3

$2.6

+674%

16k

fp8

100%

MiMo-V2.6-Pro

68.8%

$0.387

$0.783

+5%

944k

fp8

100%

GLM 5.3 Flash

67.5%

$0.15

$0.5

+406%

944k

fp4

100%

Inkling Small

67.5%

$0.45

$1.2

Only host

524k

262k

fp8

100%

GLM 5.3

67.5%

$0.5625

$2.5

+160%

131k

fp4

100%

Step 5 Preview

63.8%

$2.7

Cheapest

66k

fp8

100%

Qwen3.8 27B

63.8%

$0.15

$1.875

+242%

262k

236k

bf16

100%

MiMo-V2.6-Flash

62.5%

$0.14

$0.28

+21%

944k

fp8

100%

Nemotron 3 Ultra

60.0%

$0.4

$1.76

Cheapest

262k

16k

fp4

—

gpt-oss-120b

58.8%

$0.037

$0.17

+17%

131k

118k

bf16

100%

MiniMax M3

57.5%

$0.224

$0.88

Cheapest

524k

131k

fp8

—

gpt-oss-20b

51.2%

$0.03

$0.14

+60%

131k

118k

bf16

100%

Nemotron 3.5 Lightning

43.8%

$0.06

$0.16

+74%

262k

33k

bf16

100%

Llama 3.3 70B Instruct

42.5%

$0.1

$0.32

Cheapest

131k

16k

fp8

100%

Llama 4 Scout

41.2%

$0.08

$0.24

Cheapest

328k

295k

fp8

—

Ling 3.1 Flash

41.2%

$0.075

$0.225

Cheapest

262k

236k

fp4

100%

Phi 4

41.2%

$0.056

$0.112

Only host

16k

15k

bf16

—

Nemotron 3.5 Super VL

40.0%

$0.15

$0.5

+79%

262k

236k

fp4

100%

Qwen3.6 35B A3B

37.5%

$0.1

$0.95

+47%

262k

66k

fp8

100%

Granite 4.2 8B

36.2%

$0.06

$0.25

+30%

131k

118k

bf16

100%

Qwen3.5 397B A17B

32.5%

$0.45

+24%

262k

82k

fp8

100%

Gemma 4 31B

30.0%

$0.16

$0.32

+25%

262k

236k

fp8

—

Gemma 4 26B A4B

15.0%

$0.07

$0.34

+59%

262k

16k

fp8

100%

Ling 3.0 Flash VL

—

$0.06

$0.18

Only host

131k

33k

fp16

100%

DeepInfra: frequently asked questions

Which AI models does DeepInfra serve?

Of the models we benchmark, DeepInfra serves 26 through OpenRouter: DeepSeek V4.1 Flash 1017, DeepSeek V4 Flash Vision Exp, DeepSeek V4 Pro 0813, MiMo-V2.6-Pro, GLM 5.3 Flash, Inkling Small, GLM 5.3, Step 5 Preview, Qwen3.8 27B, MiMo-V2.6-Flash, Nemotron 3 Ultra, gpt-oss-120b, MiniMax M3, gpt-oss-20b, Nemotron 3.5 Lightning, Llama 3.3 70B Instruct, Llama 4 Scout, Ling 3.1 Flash, Phi 4, Nemotron 3.5 Super VL, Qwen3.6 35B A3B, Granite 4.2 8B, Qwen3.5 397B A17B, Gemma 4 31B, Gemma 4 26B A4B, Ling 3.0 Flash VL.

What is the most accurate model on DeepInfra?

DeepSeek V4.1 Flash 1017 by DeepSeek, with a verification score of 73.8% (59 of 80 questions correct) in our tests.

Is DeepInfra the cheapest host?

DeepInfra has the lowest list price for 7 of the 23 models it serves that other hosts also serve (DeepSeek V4 Flash Vision Exp, Step 5 Preview, Nemotron 3 Ultra, MiniMax M3, Llama 3.3 70B Instruct, Llama 4 Scout, Ling 3.1 Flash). Its median markup over the cheapest host is +24%.

How reliable is DeepInfra?

Averaged over the models it serves, DeepInfra reported 99.4% uptime in the 30 minutes before OpenRouter's catalogue was read for our run.