Consumption pricing for Private Clusters

Get started training and deploying models today via pay-as-you-go hourly or a credit-based consumption.

Gradient is designed to work seamlessly on all cloud providers and custom servers (eg virtual or bare metal). With Gradient, you only pay for what you use. Building, training, and deploying models is billed per second, often represented in hourly increments for simplicity. The compute premium is broken down by CPU and GPU workloads. There is no premium on other services used in your VPC eg storage.

Gpu instance premium =
24%
cpu instance premium =
29%
storage premium =
NONE

Gradient is designed to work seamlessly on all cloud providers and custom servers (eg virtual or bare metal). With Gradient, you only pay for what you use. Building, training, and deploying models is billed per second, often represented in hourly increments for simplicity. The compute premium is broken down by CPU and GPU workloads. There is no premium on other services used in your VPC eg storage.

public pricing coming soon
Contact Sales for pricing

Gradient is designed to work seamlessly on all cloud providers and custom servers (eg virtual or bare metal). With Gradient, you only pay for what you use. Building, training, and deploying models is billed per second, often represented in hourly increments for simplicity. The compute premium is broken down by CPU and GPU workloads. There is no premium on other services used in your VPC eg storage.

public pricing coming soon
Contact Sales for pricing

Gradient is designed to work seamlessly on all cloud providers and custom servers (eg virtual or bare metal). With Gradient, you only pay for what you use. Building, training, and deploying models is billed per second, often represented in hourly increments for simplicity. The compute premium is broken down by CPU and GPU workloads. There is no premium on other services used in your VPC eg storage.

Gpu instance premium =
12%
cpu instance premium =
15%
storage premium =
NONE

Gradient is designed to work seamlessly on all cloud providers and custom servers (eg virtual or bare metal). With Gradient, you only pay for what you use. Building, training, and deploying models is billed per second, often represented in hourly increments for simplicity. The compute premium is broken down by CPU and GPU workloads. There is no premium on other services used in your VPC eg storage.

public pricing coming soon
Contact Sales for pricing

How pricing works

Gradient pricing is very simple. Your base plan selection lets you use different Gradient tools and determines your limits (i.e. # of concurrent notebooks). You are also charged for utilization.

Total =
Subscription + Utilization

e.g. 2 : Assume you have an Essentials subscription ($129 user/mo) and you use 75 hours on a AWS P2.xlarge ($0.22/hour). Your total charge at the end of the month would be 129 + (75* 0.22) = $145.5

Subscription Pricing

View available subscription options for Self-Hosting Gradient.

See Self-Hosted subscription pricing