Administer the Platform
This section is for platform administrators. You manage what every tenant on the platform can use, from the models that run on your GPUs to the tenants and user accounts that use them.
As a platform administrator you can:
- deploy models on your GPUs and create model aliases
- share models and aliases with all tenants, or with the tenants you choose
- choose the models new RAG endpoints start with, and limit how much RAG storage each tenant can use
- create tenants and platform user accounts
- manage private networks that MCP servers are allowed to connect to
- track token usage and request counts by model and by tenant over time
You do all of this in the CM portal under Manage platform. To manage the members and access inside one tenant, see Manage the Tenant.
Start with the page that fits what you want to do:
- Quickstart: deploy and share a model: in about 15 minutes you deploy a model, give it a model alias, and share the alias with your tenants.
- Manage models and GPUs: GPU memory and model placement by node, and the platform's default models.
- Manage model aliases: stable model names you can repoint later.
- Configure a RAG endpoint: choose the models a RAG endpoint uses and who can access it.
- Set RAG storage quotas: limit how much storage a tenant or a RAG endpoint can use.
- Manage tenants across the platform: create and delete tenants, and manage their members.
- Manage platform user accounts: create, change and delete user accounts.
- Monitor platform usage: usage per tenant and per model.