Use supported cloud-based model providers

This page outlines the steps for using the integration with supported cloud-based AI model providers.

To use the APIs of supported cloud-based model providers, you must have an active account with sufficient credits or a subscription for API usage with the chosen provider. Note that account creation and credit management are handled externally and are outside the scope of this guide.

Using this feature involves transmitting data to the selected provider’s servers. We strongly recommend reviewing each provider’s privacy policy to understand their data handling and retention practices. If you handle sensitive information or require stronger data privacy guarantees, consider using a local model option instead to keep your data secure and private without external transmission.

To configure and use a cloud-hosted model provider, follow these steps:

1. Go to the Provider settings tab.

2. Select a cloud-hosted model provider from the options in the Model provider dropdown.

3. Enter the API key for the selected provider in the API key field. Refer to the selected provider’s documentation for instructions on obtaining and managing your API key.

To avoid subscription fees associated with cloud-based model providers’ APIs, consider using a local model instead. This allows you to run unlimited queries locally without relying on any external API.

1. In the Model dropdown field, type or select the exact model identifier supported by the chosen provider (e.g., gpt-4 for Open AI or gemini-3 for Google AI Gemini).

2. The Request parameters field accepts a JSON object with valid model parameters like temperature, stop, and max_tokens. For details, see the selected provider’s documentation.

3. Use the Role dropdown to select or type one of the supported roles below. This free-type field requires exact role names recognised by the system:

Role	Description
Assistant	Represents the model’s responses based on user messages. Responsible for generating replies.
System	Specifies how the model should respond. Example: “You are a helpful assistant.”
User	Represents user queries and provides context for the model’s response.