OpenAI opens Decisions API, cutting latency to 180ms

OpenAI's move follows the recent rise of TypeSafe, a startup founded by a former researcher at the frontier lab, whose Jev model created a category for fast, cheap decision models that act as super classifiers. Software developers quickly found these models could replace LLMs in some workflows at dramatically lower cost. Handa said his colleagues saw Jev and realized they had a million use cases for it, prompting the team to sprint toward a version of its own. Rather than build a specifically calibrated decision model, OpenAI reconfigured ChatGPT Luna to speed up inference and lower cost.

The competitive landscape is already crowded. Public benchmarks like the AIM System One Index and the Artificial Analysis Decision Index tend to favor D1 Ultra from Liquid and Base Lab's Elephant-1. The latest Jev model ranks seventh and third on those indices, while OpenAI's Decision Model falls eighth and fourteenth. Still, TypeSafe's offerings are for now the most widely-queried on OpenRouter, and on JevMan, a benchmark where models play PacMan, the latest Jev model tops the ranking. OpenAI's edge may lie in multi-modality, where Handa said Luna is extremely powerful and has been showing better calibrations for image classification.