GPT-6.1 Terra: Generating an Interface Image with a Markdown Prompt

A workflow proposes to engage GPT-6.1 Terra with a Markdown prompt to obtain a complete interface image, without code. The model is described as intermediate and cost-effective, focused on visual reasoning, and illustrated by a food ordering application use case.

A Markdown prompt requests a UI image, not code

The method emphasizes an explicit request: to obtain a user interface image rather than a coded prototype. The prompt is written in Markdown, with the argument that GPT models are optimized for this format. It structures expectations by listing functional details with descriptions of sections and components, a dedicated section for visual style (colors, spacing, typography), and a mention of platform conventions.

An intermediate visual model, described as useful for design

GPT-6.1 Terra is designed for visual reasoning and excels in interpreting technical drawings. Positioned between GPT Luna, which is lighter, and the larger models Sol and Astra, it is described as an intermediate and affordable solution. It is capable of assisting the entire user interface design process, from the layout exploration phase to the realization and optimization of operational interfaces.

An example: a food ordering application

The proposed workflow is organized into three steps, the first of which involves asking GPT-6.1 Terra to create a user interface design. The chosen example focuses on a food ordering application, serving as a use case to frame the prompt and specify the expected elements in the visual rendering.