Skip to main content

Developer

The default role (USER). You request virtual keys, call the gateway from your applications, and track your own spend.

Where you work: Dashboard, API Keys, Models, FinOps Analytics, Request Logs, Budgets, Documentation and Settings. The full sidebar is described in Navigating the portal.

1. Request a key​

Go to API Keys and select Request key. Pick Key ownership (You for a personal key, Service account for an app — service accounts require a team), the team and application, the environment, an access group, and a budget tier or custom budget/RPM/TPM. Your team admin, org admin or a platform admin reviews it.

2. Reveal it once and store it safely​

After approval the key is Pending reveal. Open it from API Keys and reveal it — the plaintext zea-… value is shown exactly once. Put it straight into your secret manager, never in code.

3. Point your SDK at the gateway​

Use any OpenAI SDK with base_url set to the gateway and your key as api_key. See SDK snippets. List models via GET /v1/models or the Models page.

4. Monitor usage and budgets​

Dashboard and FinOps Analytics show spend, tokens and latency; Request Logs shows individual calls; Budgets shows spend against each of your keys' budgets and any temporary increases. A 429 rate_limit_error means an RPM/TPM window is exhausted; a budget error means a budget at some level ran out.

5. Rotate or revoke​

From API Keys you can rotate (new plaintext, same settings) or revoke your own keys. Rotation takes effect on the gateway immediately.

Good to know​

  • You only see your own keys, usage and logs — no Administration group, unless you review governed spend requests (then Approvals appears).
  • If a request is pending too long, contact your team admin — they see it in Approvals.