GLM 5GLM 5
API Keys

Management API

Create, list, and revoke inference keys from automation with a management key, without granting that automation any model-inference permission at all.

Management keys are separate from inference keys and use the sk-glm5-mgmt-... prefix. They can manage keys and query Analytics, but cannot call model inference endpoints.

Create a management key in Settings → Management Keys. The plaintext secret is shown once.

List inference keys

GEThttps://glm5.app/api/v1/keys
curl https://glm5.app/api/v1/keys \
  -H "Authorization: Bearer $GLM5_MANAGEMENT_KEY"

Create an inference key

POSThttps://glm5.app/api/v1/keys
{
  "name": "production-agent",
  "rpm_limit": 60,
  "spend_limit": 300,
  "spend_reset": "daily"
}

The response includes the new inference secret once.

Read, update, disable, enable, rotate, revoke

  • GET /api/v1/keys/:id
  • PATCH /api/v1/keys/:id
  • DELETE /api/v1/keys/:id

Update fields with PATCH, or use action:

{ "action": "disable" }

Supported actions are disable, enable, and rotate. Rotation returns a replacement secret once and disables the old key. DELETE permanently revokes the key.

Management keys cannot run models

Sending a management key to /chat/completions is rejected. Use a normal sk-glm5-... inference key for model calls.

On this page

List inference keys
Create an inference key
Read, update, disable, enable, rotate, revoke