Documentation / Integrations

OpenAI

Connects OpenAI models so Helm can sample their answers during a sweep. Each connection uses one model and one request limit.

Set up

1. Open Integrations in the navigation for your role.
2. Choose Add integration, then select OpenAI.
3. Enter API key; Model name; Requests per minute.
4. Choose Connect. Helm tests the connection before it saves the integration.

If the test fails, you can choose Save without verifying. The integration is then saved with an error status until you fix it.

What the screen shows

The form asks for an API key, the exact model identifier and a request limit, which defaults to 600 per minute. The key field has a show control. When you choose Connect, Helm checks the key with OpenAI before it saves the connection.

Notes

• Enter the exact model identifier, because that name is what the provider receives.
• Lower the request limit if the same key is used by other tools in your organization.

Last updated 26 October 2026