Skip to main content
Every call through the LLM Gateway is paid one of three ways, called license models: Admins choose which ones people may use in Policies (License model). For each call, the gateway decides in this order:

​
1. A person’s own Claude plan

Using a person’s own Claude plan is available to organizations where xpander has enabled it. Contact xpander to turn it on.
A call runs on the person’s own Claude plan, including any extra usage the plan has, when all of these hold:
  • the person uses Claude Code, signed in with their Claude plan, and sends their xpander key in the x-xpander-api-key header
  • the model is a Claude model from Anthropic
  • Bring your own subscription is on for the person’s organization, group and agent
If any condition is not met, the call is refused with a message that names what is missing. The call is never silently moved to company models.

​
When the plan hits its limit

When someone’s plan hits its limit in Policies decides what happens until the plan resets: When your organization falls back, each person can still choose to pause their own calls: When my Claude plan hits its limit in AI models & keys offers Continue on company models or Pause. When your organization pauses, that choice is locked for everyone with Your admin set calls to pause when a plan hits its limit. While a plan is limited, AI models & keys says so, for example Your Claude plan’s usage limit is reached until Thu, Oct 8, 8:00 AM. Calls use company models until then.

​
2. Company models

Every other call, and a call a limited plan falls back from, runs on company models: a key your organization stored, or xpander’s key. Default license model in Policies decides which. With Bring your own key, the stored key used is the first one assigned in this order: the agent making the call, the person, the person’s primary group, then the organization default for that provider (Whose key pays).

​
Cloud, Hybrid and Air-Gapped

On a Hybrid or Air-Gapped install, add a key for every provider your people use. A call for a provider with no key is refused. When some of your agents run on your install and others in xpander cloud, the agents in xpander cloud may still use Included with xpander.

​
See who paid

Open a request in Logs. License model shows how it was paid, its Gateway decisions show whether the plan, a stored key or xpander’s key served it and why, and the License model filter lists the requests paid each way. In Usage, the tiles split spend into Subscription: xpander credits, Subscription: your provider keys and Subscription: personal Claude and ChatGPT plans.