API host · catalogue checked 27 Oct 2026

Phala: models, prices and uptime

Phala serves 9 of the 78 models we benchmark through OpenRouter. Prices are OpenRouter's list prices per million tokens. We did not measure speed per host.

Models served9of 78 we test

Cheapest host for0of 9 shared models

Median premium+297%over the cheapest host

Uptime100%last 30 min, average

Most accurate models on Phala

Verification score of each model this host serves

1DeepSeek V4.1 Flash 1017DeepSeek73.8%
2DeepSeek V4 Pro 0813DeepSeek71.2%
3GLM 5.3 FlashZ.ai67.5%
4GLM 5.3Z.ai67.5%
5Qwen3.8 27BQwen63.8%
6Hy3Tencent58.8%
7gpt-oss-120bOpenAI58.8%
8Qwen3.6 35B A3BQwen37.5%
9Qwen3.5 397B A17BQwen32.5%

Models on Phala

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%	$0.2	$1.2	+297%	30	1M	393k	—	100%
DeepSeek V4 Pro 0813	71.2%	$0.957	$2.8776	+585%	23	1M	393k	—	100%
GLM 5.3 Flash	67.5%	$0.1275	$0.425	+330%	34	1M	131k	nvfp4	100%
GLM 5.3	67.5%	$0.84	$2.64	+221%	37	1M	944k	—	100%
Qwen3.8 27B	63.8%	$0.2	$2.5	+356%	21	1M	262k	—	100%
Hy3	58.8%	$0.15	$0.64	+89%	5	262k	236k	—	100%
gpt-oss-120b	58.8%	$0.15	$0.6	+337%	23	131k	33k	—	100%
Qwen3.6 35B A3B	37.5%	$0.2	$1.27	+120%	10	262k	236k	—	100%
Qwen3.5 397B A17B	32.5%	$0.55	$3.5	+47%	10	262k	236k	—	100%

All API hosts →

Phala: frequently asked questions

Which AI models does Phala serve?: Of the models we benchmark, Phala serves 9 through OpenRouter: DeepSeek V4.1 Flash 1017, DeepSeek V4 Pro 0813, GLM 5.3 Flash, GLM 5.3, Qwen3.8 27B, Hy3, gpt-oss-120b, Qwen3.6 35B A3B, Qwen3.5 397B A17B.
What is the most accurate model on Phala?: DeepSeek V4.1 Flash 1017 by DeepSeek, with a verification score of 73.8% (59 of 80 questions correct) in our tests.
Is Phala the cheapest host?: Phala has the lowest list price for 0 of the 9 models it serves that other hosts also serve. Its median markup over the cheapest host is +297%.
How reliable is Phala?: Averaged over the models it serves, Phala reported 100% 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%

$0.2

$1.2

+297%

393k

—

100%

DeepSeek V4 Pro 0813

71.2%

$0.957

$2.8776

+585%

393k

—

100%

GLM 5.3 Flash

67.5%

$0.1275

$0.425

+330%

131k

nvfp4

100%

GLM 5.3

67.5%

$0.84

$2.64

+221%

944k

—

100%

Qwen3.8 27B

63.8%

$0.2

$2.5

+356%

262k

—

100%

Hy3

58.8%

$0.15

$0.64

+89%

262k

236k

—

100%

gpt-oss-120b

58.8%

$0.15

$0.6

+337%

131k

33k

—

100%

Qwen3.6 35B A3B

37.5%

$0.2

$1.27

+120%

262k

236k

—

100%

Qwen3.5 397B A17B

32.5%

$0.55

$3.5

+47%

262k

236k

—

100%

Phala: frequently asked questions

Which AI models does Phala serve?

Of the models we benchmark, Phala serves 9 through OpenRouter: DeepSeek V4.1 Flash 1017, DeepSeek V4 Pro 0813, GLM 5.3 Flash, GLM 5.3, Qwen3.8 27B, Hy3, gpt-oss-120b, Qwen3.6 35B A3B, Qwen3.5 397B A17B.

What is the most accurate model on Phala?

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

Is Phala the cheapest host?

Phala has the lowest list price for 0 of the 9 models it serves that other hosts also serve. Its median markup over the cheapest host is +297%.

How reliable is Phala?

Averaged over the models it serves, Phala reported 100% uptime in the 30 minutes before OpenRouter's catalogue was read for our run.