Manage Deployments
Listing Deployments
Dashboard
Navigate to Deployments to see all active and historical deployments for your account (or organization, if you are in one). Use the filter tabs to view by status.
API
Authentication uses the session cookie set when you sign in (see Authentication).
import requests
COOKIES = {"session": "<your-session-cookie>"}
r = requests.get(
"https://api.xinference.co/api/v1/model-deployments",
cookies=COOKIES,
params={"limit": 100, "offset": 0},
)
data = r.json()
print(f"Total: {data['total']}")
for d in data["items"]:
print(d["id"], d["model_name"], d["status"])
Query parameters:
| Parameter | Default | Description |
|---|---|---|
limit |
100 | Number of results to return (1–500) |
offset |
0 | Pagination offset (≥ 0) |
Retrieving a Deployment
r = requests.get(
"https://api.xinference.co/api/v1/model-deployments/dep_abc123",
cookies=COOKIES,
)
deployment = r.json()
Terminating a Deployment
Send a DELETE request to stop a deployment. This initiates termination and returns the updated deployment object.
requests.delete(
"https://api.xinference.co/api/v1/model-deployments/dep_abc123",
cookies=COOKIES,
)
Termination is asynchronous. The deployment status transitions:
terminating_model → releasing_instance → terminated
Poll the status endpoint or watch the dashboard to confirm termination.
Cost Control
Terminate deployments when not in use. With per-second billing, even a few hours of idle time can add up for large GPU instances.
Deployment Retention
Terminated deployment records are retained indefinitely for billing audit purposes. They appear in the deployment list with status terminated and are not billed.
Concurrent Deployments
By default, each account can run a limited number of simultaneous deployments. If you need higher concurrency, contact support or check your organization's quota in Organization → Settings → Quotas.
Organization-Level View
Owners and Admins can see all deployments across their organization — not just their own — from the Organization → Deployments page. This view supports filtering by user, model, and status.