Connecting to Models

You'll need at least one model from a supported provider to use Promptscape

Supported Model Providers

The following model providers are supported by Promptscape.

Provider	Description
Ollama	Local models running on your machine using Ollama
OpenAI	Models hosted on OpenAI's endpoints
Azure	Models hosted in Azure, in your own deployment of the Azure OpenAI Service
DeepSeek	Deepseek chat & reasoner (R1) models, on the DeepSeek Platform

You can add any number of models to Promptscape, which is particularly convenient if you have multiple use cases, or want to test an existing prompt on a new model or provider.

Adding Models

Use the Promptscape: Create a Model command to add your first model to Promptscape. The extension will guide you through the steps of providing the right information for the provider you select.

Authentication & Authorization

Certain providers may require authentication when using them - notably Azure & OpenAI deployments. Promptscape supports key-based authentication, and these keys are securely stored in the VSCode secrets vault. The keys are removed permanently from the vault when the associated models are deleted. Nonetheless, the use of production deployments and their associated keys is not recommended, and in the case of Azure, key-based authentication is strongly discouraged in production, as a rule.

Model Commands

Promptscape: Create a Model: This command initiates the onboarding for a new model. You will be asked to provide the requested information through the VSCode input box at the center top of the window.

Promptscape: Set Model: At all times, you should have one model activate. The extension uses this active model when you experiment with your prompts through the Studio. To activate a model, or change which model is active , use this command, or simply click on on the desired model in the Available Models view in the Explorer.

Promptscape: Delete Model: Permanently deletes the model and its associated configurations & secrets from the workspace.