AI Model Settings

AI Model Settings tells PromptHub which model service to use, where its endpoint is, and which model should handle each supported feature.

You do not need to configure every provider. Start with one reliable text model, test the connection, then add vision or image capabilities only when a workflow needs them.

When to Open This Page

• Before using prompt testing, AI editing, translation, or image-related features.
• After a credential expires or a provider changes its endpoint.
• When you want one feature to use a different model.
• When a team gives you a private model endpoint.

Configuration Flow

1. Open AI Model Settings.
2. Choose a provider or a compatible custom service.
3. Enter the credential and optional base URL supplied by that service.
4. Test the connection before relying on it.
5. Add or fetch models supported by the endpoint.
6. Select a default model for each feature you intend to use.

Choosing a Default

Workflow	Practical Starting Point
Writing, summarizing, translation	A stable text model
Code review and long documents	A model with enough context for the input
Image prompt analysis	A model that accepts visual input
Repeated prompt comparison	A predictable model with acceptable cost and latency

Model names and availability come from the provider or custom endpoint, so they can change independently of PromptHub.

Troubleshooting

Message	Likely Cause	What to Check
No model configured	The feature has no default route	Select a default model for that feature
Authentication failed	Invalid credential or missing provider permission	Re-enter the credential and check account access
Connection timed out	Network, proxy, or endpoint issue	Check the network and base URL, then retry
504	The model or gateway did not answer in time	Retry later or choose a more reliable endpoint

Security

Store model credentials only in the product’s local configuration or a trusted environment-variable boundary. Never place them in prompts, skills, rules, screenshots, or public documentation.