> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scitix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# API Keys

API keys authenticate programmatic requests to the ScitiX Model Inference API. Use **one key per application or environment** so development, staging, and production traffic can be managed independently.

<img src="https://mintcdn.com/scitix-adfc65c2/zz3UuyM3SCvVbeP9/model-inference/media/api-keys.png?fit=max&auto=format&n=zz3UuyM3SCvVbeP9&q=85&s=ed2ab34dd2314779b77bbf417cacc990" alt="API Keys" width="3822" height="1792" data-path="model-inference/media/api-keys.png" />

## Create a Key

Go to the [API Keys](https://console.scitix.ai/model-inference/api_keys) page and click **Create API Key**. Give it a recognizable name, then copy the key and store it in an environment variable instead of hard-coding it:

```bash theme={null}
export API_KEY="YOUR_API_KEY"
```

<Warning>
  Treat API keys like passwords. Never commit them to Git or expose them in client-side code.
</Warning>

## Control Usage

Each key supports quotas and rate limits. In shared environments, experiments, or third-party integrations, these settings help you define clear boundaries for cost and throughput.

* **Quota limit settings**: Use a **Monthly Quota** for the current month or a **Total Quota** for lifetime spend.
* **Rate limit settings**: Use a **TPM Limit** (tokens per minute) and an **RPM Limit** (requests per minute) to control throughput.

Leave a field empty or set it to `0` for unlimited.

## Manage Keys

Each key provides management actions in the key list:

* **Edit**: Modify the quota limit or rate limit.
* **Pause**: Temporarily block a key without deleting it. This is the safest first step when investigating unexpected usage, because you can re-enable the key later.
* **Delete**: Permanently revoke a key. This cannot be undone. Update any application that still depends on the key before deleting it.

## Usage Recommendations

The key list shows each key's **status**, **last-used time**, and **monthly token usage**. If traffic looks unusual, review your usage trends first, then tighten quotas or rate limits. You can also pause the key while investigating.