Overview
Manage your team's access to GPU resources and control costs with resource and pricing policies.
Resource Policies
Resource policies help you manage how your teams can access and use your GPU resources.
You can define access levels, set usage limits, and control deployment locations. Manage which regions are available for deployment and set resource limits globally, regionally, or by resource group. You can also manage access to individual GPU pools.
Each team must have one resource policy. Policies cannot be unassigned, but can be edited or replaced.
Pricing Policies
Pricing policies govern resource consumption costs for teams. Pricing varies by resource type.
You can set up monthly costs regardless of actual usage. You can also define inclusive amounts in a specific currency for plans with included resources. Each pricing policy requires a currency assignment, which is locked after you assign the policy to a team. Usage costs are tracked and accessible through the API to integrate with your billing systems.
Access Levels and Types
You can manage access at three levels: Global, Regional, and Resource Group. For example, a global restriction of 100 GPU hours would limit all teams to a maximum of 100 GPU hours across all regions and resource groups.
Global: Restrictions apply to all resources for the assigned teams.
Regional: Restrictions apply to specific regions you choose. Regional limits cannot exceed global limits.
Resource Groups: Restrictions apply to a subset of resources within a region. Resource group limits cannot exceed regional or global limits.
For each level, you can set one of three access types:
No Access: The team cannot use the specified resource(s).
Restricted: Set a maximum usage value for the resource(s).
Unlimited: The team has unrestricted usage of the resource(s).
Resource Pricing
Resource pricing is determined by the following metrics. These costs are factored into your overall billing depending on the chosen pricing policy.
CPU: Price per core per hour.
RAM: Price per GB per hour.
Disk Storage: Price per GB per hour.
Bandwidth (In/Out): Price per GB.
IP Address: Price per public IP per hour.
GPU Card: Price per hour.
Billable Service: Price per hour.
GPUaaS Pool Pricing
GPUaaS pool pricing is based on the following. These costs are added to your resource costs based on your usage.
Per Subscription / Hour
vRAM: Price per GB.
TFLOPS: Price per TFLOPS.
Policy Assignment
Each team must have one resource policy and one pricing policy.
Full Administrators choose policies when creating a team via UI. Normal users are assigned default policies when creating a team via UI. API-created teams inherit default policies if none are specified.
Combined Policy Functionality
Combine resource and pricing policies to create flexible billing scenarios:
No Commitment: Pay as you go, based on used resources. You are billed for what your teams use.
Commitment: Pay a fixed monthly fee for a set amount of resources. "Overage" charges apply for usage beyond the committed amount. For example, if your commitment includes 100 GPU hours and your team uses 110, you'll be billed the commitment fee plus the hourly rate for the extra 10 hours.
Prepaid Package with Inclusive Resources: Purchase a package with a defined set of resources for a fixed price. Overage fees apply for usage beyond the included resources. For example, a package might include 100 GPU hours. If your team uses 120 hours, you will be billed for the additional usage.
Prepaid Package with Specific Resources only: A prepaid package providing access to a specific set of resources. Usage is limited to these resources.