OpenAI launches Decisions Playground and expands web indexing

OpenAI has released the Decisions Playground, an interactive testing environment designed to make its Decisions API more accessible to software developers. The tool allows users to test the API's ability to perform structured analysis and classification tasks for text and image inputs. The API provides responses in three formats: as a probability (Predicate), a selection (Choice), or a numerical scale (Score).

Currently powered by the gpt-6-luna model, the Decisions API is reported to operate up to ten times faster than the same model via the Responses API. The base pricing is set at $0.10 per one million input tokens, with output tokens provided free of charge.

In a separate development, a research pilot conducted with Cloudflare has enabled OpenAI to discover and index hundreds of millions of previously unknown web pages. By utilizing Cloudflare’s network signals—such as page update frequency and traffic quality—OpenAI’s search crawler can more efficiently identify valuable content and avoid unnecessary requests to non-existent or redundant files. The pilot also includes testing for signals that flag potentially harmful pages, such as fake login forms.