> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.runtype.com/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.runtype.com/_mcp/server. # Platform keys vs. bring your own key (BYOK) Runtype runs AI models with either Runtype platform keys or provider credentials that you connect. Platform keys are included with every plan. Bring your own key (BYOK) gives you dedicated provider limits, direct provider billing, and access to more of each provider's models. ## Platform keys Platform keys are Runtype-managed API credentials that are included with every plan. They are available when you sign up, so you can build Flows and Agents without creating provider accounts. Runtype tracks platform-key spend for you. Platform keys work well for prototypes, demos, and exploring the platform. They cover a curated selection of models from providers such as OpenAI, Anthropic, Google, xAI, and Mixlayer. For the full list, see [Available AI models](/user-guide/settings/available-ai-models). Consider these trade-offs before you rely on platform keys in production: * Platform keys use shared Runtype capacity, so provider throttling can affect response times during peak usage. * Platform-key usage draws on your Runtype credit. If the credit runs out, models that use platform keys stop working until you add credit or a payment method. Models that use your own provider credential keep working. * OpenAI Web Search, Anthropic Web Search, and xAI search carry a small per-invocation fee on top of token costs. With your own provider credential, you pay the provider directly. For details, see [Built-in tools](/user-guide/tools/built-in-tools). On **Settings > Models**, a model that runs on a platform key shows a **Platform Key** badge. ## Bring your own key (BYOK) BYOK connects your own provider account to Runtype. Requests use that account's limits, and the provider bills you directly. Depending on the provider, you connect an API key, a service-account credential, or AWS credentials. The following table lists the providers that you can connect and whether each one also offers a platform key: | Provider | Platform key available | | -------------------------- | ---------------------- | | OpenAI | Yes | | Anthropic | Yes | | Google | Yes | | xAI | Yes | | Mixlayer | Yes | | Vercel AI Gateway | Yes | | Tinfoil | No | | Amazon Bedrock | No | | Vertex AI | No | | Vertex AI (Claude) | No | | Custom inference endpoints | No | Custom inference endpoints are servers that speak the OpenAI Chat Completions or Responses API, such as Groq and Fireworks, and self-hosted vLLM servers. You can use platform keys and BYOK at the same time. For example, use your OpenAI credential for production Flows and platform keys for Anthropic models that you are testing. ### Before you set up BYOK To add a provider connection, you need the following: * A Team plan or higher, with a payment method on file. If a provider shows **Add a payment method**, open **Settings > Billing** and add one. * In an organization, the organization admin role. Other members must ask an admin to add or change a connection. * A credential from the provider, such as an API key that you create in the provider's dashboard. For the steps and each provider's credential requirements, see [Connecting AI model providers](/user-guide/integrations/connecting-ai-model-providers). ## Organization connections In an organization, admins save provider connections that the whole team shares. The organization owns each connection. Runtype records which member added or changed it. Saving a connection doesn't make Runtype use it. To use a connection for every Flow and Agent in the organization, an admin selects it for the provider. The selection covers runs that any member starts, plus scheduled runs, API calls, and Surfaces. You don't need to update individual Flows or Agents. To select a connection, follow these steps: 1. On **Settings > Models**, select the **Providers** tab. 2. Next to the provider, click **Manage connections**. 3. Next to the connection, click **Use for the organization** (the star). 4. In the confirmation dialog, click **Use this connection**. The selected connection shows a **Used organization-wide** badge. Existing Flows and Agents use it on their next run. To select a connection through the API, send `PUT /v1/provider-keys/organization-policy`. Through MCP, call `select_organization_provider_credential`. For an example request, see [Connecting AI model providers](/user-guide/integrations/connecting-ai-model-providers). ### Replace or remove a selected connection After you select a connection, Runtype doesn't switch to the platform key if that connection stops working. Before you revoke or delete the selected connection, select a replacement: * To return to Runtype's platform key for OpenAI, Anthropic, Google, xAI, Mixlayer, or Vercel AI Gateway, open **Manage connections**, click **Use Platform for organization**, and then click **Use the plan key**. Through the API, send `"selection": "platform"`. * For a provider without a platform key, select another connection first. To rotate the credential of the selected connection, edit the connection. The connection stays selected. ## Compare platform keys and BYOK The following table compares the two options: | Category | Platform keys | BYOK | | ----------------- | ------------------------------------- | --------------------------------------------------------- | | Setup | Available when you sign up | Add and save a provider credential | | Rate limits | Shared Runtype capacity | Dedicated provider limits | | Model access | Curated selection of supported models | Supported models from your provider | | Billing | Runtype tracks platform-key spend | The provider bills you directly | | Best for | Starting and prototyping | Production workloads, provider limits, and direct billing | | Plan availability | All plans | Team plan and higher | ## Troubleshooting If a provider connection request fails, check the status code: * `402` with code `BYOK_NOT_AVAILABLE`: Your plan doesn't include BYOK. Upgrade to the Team plan or higher, or add a payment method. * `403`: You aren't an organization admin, or your API key lacks the `INTEGRATIONS:WRITE` permission. * `422` when you revoke, delete, or change a selection: The provider would have no usable credential. Select another connection or the platform key first. ## Next steps Continue with these guides: * [Connecting AI model providers](/user-guide/integrations/connecting-ai-model-providers): add provider credentials and enable models * [Available AI models](/user-guide/settings/available-ai-models): review models available with platform keys and BYOK * [Rate limits and usage](/user-guide/troubleshooting-faq/rate-limits-and-usage): understand execution, provider, and spend limits * [Model routing and data flow](/user-guide/settings/model-routing-and-data-flow): see which provider serves each model and who can see your prompts ### Provider changes with BYOK-only enabled When **Require your own model keys** is enabled, select another organization connection before removing the selected one. Selecting a platform key is refused. To return to platform credentials, an admin must first turn off that restriction in **Settings > Models**.