AI providers and models

Connect the providers your organization uses and choose which models to offer.

When AI is needed

• AI generates and edits tests. You can configure it after first signup; it is not required to claim or finish installation setup.

Add a provider

• Open Settings → Administration → Model providers. Enter a provider ID, type, display name, endpoint and API key as required.
• Save the provider, test the connection, refresh its model list, and enable the models your team may use. Stored API keys are never returned to the browser.

Provider compatibility

• Use a supported provider type or an OpenAI-compatible endpoint. Choose a model that supports the capabilities your testing workflow needs.
• Confirm a small generation succeeds before offering a new model to the whole team.

Private networks

• For an offline installation, use a model endpoint reachable from the TestVibe host and provide required model/runtime assets through your deployment process.
• A provider running on the operator’s laptop is not reachable as localhost from a container or appliance. Use the host-visible address.

Choose project models

• Enable models in Administration, then select the appropriate model in the project’s model settings.
• Refreshing discovery does not intentionally re-enable models you disabled.

Connection problems

• Use Test connection and confirm the endpoint, authentication and network route. Check private CA and proxy requirements in your environment.
• Do not clear a stored key when you only need to change the endpoint.

Capacity and provider charges

• Your AI provider may bill usage independently of the TestVibe license. Check provider quotas and monitor usage in Administration → Operations and activity.

Existing configuration

• Previously configured provider values remain on upgrade. Host-managed values take precedence over editable installation settings.
• Keep your existing encryption key when migrating or restoring provider credentials.