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

# Bring Your Own Key (BYOK)

> Use your own LLM API key with Stewrd. You control model choice and costs.

## What is BYOK?

Stewrd is managed AI agent infrastructure — orchestration, browser automation, file storage, and sessions. With BYOK, you bring your own LLM API key so you control which model runs and how much inference costs.

Stewrd charges only for infrastructure. Your LLM costs go directly to your provider.

## Supported Providers

| Provider                              | Status    | Base URL                       |
| :------------------------------------ | :-------- | :----------------------------- |
| [OpenAI](https://platform.openai.com) | Supported | `https://api.openai.com/v1`    |
| [OpenRouter](https://openrouter.ai)   | Supported | `https://openrouter.ai/api/v1` |

**OpenAI** — Use your OpenAI API key directly for access to GPT-4o, GPT-4o mini, o1, and other OpenAI models.

**OpenRouter** — Access hundreds of models (GPT-4o, Claude, Llama, Mixtral, and more) through a single API key.

## How to Configure

### 1. Get a provider API key

* **OpenAI:** Sign up at [platform.openai.com](https://platform.openai.com) and create an API key under API Keys.
* **OpenRouter:** Sign up at [openrouter.ai](https://openrouter.ai) and create an API key.

### 2. Configure in your dashboard

Go to your [project settings](https://stewrd.dev/dashboard), select your project, and click **Edit** under LLM Configuration:

* **Provider**: Select OpenAI or OpenRouter
* **API Key**: Paste your provider API key
* **Model**: Choose a model (see table below)

### 3. Send requests as normal

Your API requests work exactly the same — Stewrd uses your key to call the LLM provider on your behalf.

```bash theme={null}
curl https://api.stewrd.dev/v1/agent \
  -H "Authorization: Bearer sk-stw_your_key" \
  -H "Content-Type: application/json" \
  -d '{"message": "Summarize the latest AI news"}'
```

## Model Selection

You can use any model available on your provider. Some popular choices:

| Model                               | Provider   | Best For                                |
| :---------------------------------- | :--------- | :-------------------------------------- |
| `gpt-4o`                            | OpenAI     | High quality reasoning                  |
| `gpt-4o-mini`                       | OpenAI     | Fast, cost-effective                    |
| `o1`                                | OpenAI     | Complex reasoning tasks                 |
| `moonshotai/kimi-k2.5`              | OpenRouter | General purpose, good balance           |
| `openai/gpt-4o`                     | OpenRouter | High quality reasoning (via OpenRouter) |
| `anthropic/claude-sonnet-4`         | OpenRouter | Nuanced analysis, coding                |
| `meta-llama/llama-3.1-70b-instruct` | OpenRouter | Cost-effective, fast                    |

## Security

Your provider API key is protected with multiple layers:

* **Encrypted at rest** using AES-256-GCM
* **Never logged** in any application or access logs
* **Never returned** in any API response (only a `has_provider_key: true/false` flag)
* **Transmitted securely** over HTTPS between Stewrd services
* **Scoped per project** — each project can have its own key and model

<Warning>
  If your provider key is compromised, rotate it in your provider's dashboard and update it in your Stewrd project settings.
</Warning>

## Without BYOK

If you don't configure a provider key, Stewrd falls back to its default model configuration. This may have limited availability — we recommend configuring your own key for production use.