Checking OpenAI
Credit$—

Agent

Agent configuration

Define instructions, tools, model behavior, and execution limits.

Draft

Identity

Basic metadata used in the dashboard and API.

Instructions

Sent to the model at the beginning of every run.

Tools

Enable only the capabilities this agent requires.

Web searchSearch current sources on the public web.
Code interpreterRun code and work with generated files.

Playground

Test a deployment

Inspect streamed output, tool activity, approvals, and cost.

Agent session

No messages yet

Send a prompt to start a stateful agent session.

⌘↵ to run

Deployments

API access

Create keys and call pinned agent versions from your product.

Quick start

Start a session, then create a metered run.

REST
Select or deploy an agent to see the API example.

API keys

Keys are hashed and shown only once.

Deployments

Each release points to a frozen agent version.

AgentVersionEndpointModelStatusCreated

Usage

Usage and billing

Token consumption and settled charges across all runs.

Pricing —
Available credit$—Prepaid balance
Total charged$—Settled usage
Input tokens—Across all runs
Output tokens—Across all runs

Recent runs

Reservation, actual cost, and terminal status.

RunModelStatusTokensChargedStarted

Control-plane access

Enter the production admin token. It remains in this browser tab.

Create API key

The secret value will only be shown once.

Revoke API key

Requests using this key will stop working immediately.