# API Quickstart

Make your first authenticated request to Redbelt by listing the models available
to your API key. This guide uses Python's standard library.

## Before you start

You need a Redbelt account, an organization with API access, and Python 3.10 or
later. Creating a key requires permission to edit your organization's admin settings.
Ask your administrator for access if **Developer** or **Create key** is unavailable.

## Create your key

1. Sign in to [Redbelt](https://redbelt.ai) and open **Settings**.
2. Select **Developer** under **Admin**.
3. In **API keys**, click **Create key**.
4. Choose **Service account** for an integration identity, or **You** to make
   requests as your user. Enter a name you will recognize later.
5. For a service account, review **Auto share resources with creator**.
   This shares resources created by that key with you; it does not grant the
   service account access to all your existing private resources.
6. Click **Create secret key**. If prompted, use **Sign in again** and repeat
   the creation step.
7. In **Save your key**, click **Copy**, store it securely, and click **Done**.
   The full key is displayed only once.

See [Create and manage API keys](/api/keys/) for ownership and revocation details.

![Current Create API Key form, showing ownership, name, and resource sharing options.](/screenshots/current/create-api-key.jpg)

## Set your environment

Set **REDBELT_API_KEY** in the environment where Python will run. In a terminal,
this reads the value without displaying it or putting it in a literal shell command:

```bash
read -rs REDBELT_API_KEY
export REDBELT_API_KEY
```

Paste your key and press Enter. Keep it out of source files and browser code.
For a self-hosted deployment, also set **REDBELT_API_URL** to its API base URL,
including the **/api** path.

## List available models

Save this as **list_models.py**:

```python
"""List models available to the API key identity."""

import json
import os
from urllib.request import Request, urlopen

api_url = os.environ.get("REDBELT_API_URL", "https://redbelt.ai/api").rstrip("/")
request = Request(
    f"{api_url}/v1/models",
    headers={"Authorization": f"Bearer {os.environ['REDBELT_API_KEY']}"},
)
with urlopen(request, timeout=30) as response:
    models = json.load(response)
print(json.dumps(models, indent=2))

```

Run it:

```bash
python3 list_models.py
```

A successful request returns HTTP **200** and a JSON array. The entries depend on
your organization's configured providers and your access. An empty array is a
valid response; it does not prove a model provider is configured.

## If the request fails

* **401:** check that the key is present, belongs to this deployment, and has not
  been revoked.
* **403:** confirm the API key identity has the required organization access.
* A connection or hostname error: check the API base URL and your network.
* Missing models: ask your administrator to review model/provider configuration.

Next, [make a direct AI request](/api/direct-inference/),
[run an agent](/chat_agents/), or [run a workflow](/api/workflow-runs/).
