AI Providers
Configuring AI provider API keys for use in agent nodes.
AI provider keys let your workflows use large language models in Agent nodes with your own API keys. Keys are configured at the organization level (on the Resources page) and enabled per workspace.
You don't need a key to start. Every plan includes monthly AI credits, and Agent nodes run on those credits out of the box. Different models consume credits at different rates, more capable models use more per call. Adding your own provider key means agent calls are billed directly by the provider and no longer count against the credit allowance.
Supported Providers
| Provider | Models |
|---|---|
| Anthropic | claude-opus-4-6, claude-sonnet-4-6, claude-opus-4-5, claude-sonnet-4-5, claude-haiku-4-5 |
| OpenAI | gpt-5.4, gpt-5.3, gpt-5.2, gpt-5, gpt-5-mini, o4-mini |
| gemini-2.5-pro, gemini-2.5-flash, gemini-2.0-flash |
Adding a Provider Key
- Navigate to Resources > AI Providers.
- Click Add Provider.
- Select the provider type (Anthropic, OpenAI, or Google).
- Enter a display name for the key (e.g., "Production Anthropic Key").
- Paste your API key.
- Pick the workspaces where the key should be available.
- Click Save.
The API key is encrypted before being stored. The full key is never displayed again after saving.
The number of provider keys you can configure depends on your plan.
Managing Keys
Edit
You can update the display name of an existing key or toggle it active/inactive. Inactive keys are not available for use in workflows.
Delete
Deleting a key removes it permanently. Any workflows using this key will fail on their next execution if no other valid key is available in the workspace.
Enabling in Workspaces
Provider keys must be enabled per workspace before workflows there can use them. Either pick the workspaces when creating the key, or later:
- Open the workspace's settings (
/workspaces/<workspace>/settings). - Go to the Enabled resources tab.
- Under AI Providers, check the keys you want to make available.
- Save.
Only enabled providers appear in the Agent node model dropdown within the workflow editor.
Runtime Behavior
When a workflow executes, the engine loads the AI provider keys enabled for the workflow's workspace. If the selected model has no matching enabled key, the agent call falls back to the platform key and is charged against your plan's monthly AI credits. There is no fallback to environment variables or hidden defaults.