Decision Studio Lets Users Test Decision Models Without Writing API Requests

LLM Gateway has introduced Decision Studio, a new window in Lounge for trying decision models that return probabilities rather than generated text. Previously, testing these models, which are served through an OpenAI-compatible /v1/decisions endpoint, required hand-writing JSON requests and reading raw responses. The studio lets users paste an input such as a support ticket or review, then add up to ten questions of yes/no, pick-one, or rating types. Results are displayed as cards showing the headline answer and per-option probabilities, with options to copy the equivalent cURL request or view raw JSON. Six presets load sample inputs and questions to demonstrate the different answer formats.