API quickstart

Make your first API request.

Use an API key to call a model from Python.

1. Sign in

Open the console with the email your ThirdBrain Labs contact added. Select your organization if prompted.

2. Create an API key

In Organization → API keys, select Create key. Copy the key when it appears and set it as TBL_API_KEY in your environment or secret manager.

Only organization Admins can create and revoke keys. Admins manage all keys in their organization. If you do not see Create key, ask your Admin for API access. Keep your key private.

3. Send a request

Example request using the OpenAI Python SDK:

Python example
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["TBL_API_KEY"],
    base_url="https://api.dev.thirdbrainlabs.ai/v1",
    max_retries=0,
)

response = client.chat.completions.create(
    model="openai/gpt-6-luna",
    messages=[{"role": "user", "content": "Say hello."}],
    max_completion_tokens=64,
    reasoning_effort="none",
)
print(response.choices[0].message.content)

After your first request

Browse available models to copy another model ID. You can also use GET /v1/models with your key.

Check usage and spend in your dashboard before increasing traffic. API requests can incur charges.

Other APIs and streaming

OpenAI-compatible SDKs use https://api.dev.thirdbrainlabs.ai/v1. The Anthropic SDK uses https://api.dev.thirdbrainlabs.ai without the suffix.

Responses requires store=False; Anthropic Messages requires max_tokens. For streaming, read through the final event. Keep automatic retries disabled: a timed-out request may already have run.

Default limits

Your organization’s key defaults appear above the API keys in your dashboard. Standard keys expire after 30 days and allow 30 requests per minute with a burst of 4. An approved organization policy can remove the gateway key rate cap and automatic expiry. Organizations allow 4 concurrent requests by default, including open streams, unless a separate organization policy changes this.

Provider and gateway infrastructure limits still apply. Usage is recorded for invoicing afterward. With a monthly spend limit, new requests stop at the cutoff shown in your dashboard based on settled charges. Requests already running finish and can take spend above the limit. Missing usage stays pending for reconciliation.

Need help?

If access was added after you signed in, open Switch or refresh access, choose your organization, and select Select and refresh access.

Contact the person who onboarded you with the request ID from x-request-id, model, and error. Never include your API key or private prompt.