OpenAI has made its Decisions API available to all customers, entering the market for fast, low-cost decision models that recently made TypeSafe, a startup founded by a former researcher at the frontier lab, a $7.6 billion company. TypeSafe‘s Jev model defined the category: given a set of options, these models act as super classifiers, quickly making a choice and returning a confidence rating. Developers found that such decision models, sometimes called System One models, could replace LLMs in many workflows at dramatically lower cost and higher speed.

OpenAI‘s API product lead, Nikunj Handa, said his colleagues saw Jev and concluded “we have like a million use cases for this.” Rather than building a specially calibrated decision model, the team took ChatGPT Luna and reconfigured it to speed up inference and cut cost. “The big unlock with Decisions API has been the speed, shrinking things from one to three seconds to 100, 120, 180 milliseconds,” Handa said. “In the future, we may sort of have specific training to make our models even better in these ways. But right now, it’s just plain old Luna that gets the job done.” The API was announced at OpenAI‘s developer day on September 29 and opened to all customers on Thursday. The public version adds voice and visual modalities, and OpenAI targeted inference resources geographically to lower latency in different regions.

Customer trials with Box, Datadog, Ramp, and Speak show the latency gains. Andrew Hsu, CTO of language-tutor builder Speak, said the Decisions API cut the time between actions from 1.4 seconds to 0.13 seconds. OpenAI also used the product to improve responsiveness in its Dots voice assistant.

Competition in the category is heating up. It is not clear what makes a good decision model beyond “it works for you”: intelligence, cost, calibration, and speed trade-offs differ by workload. TypeSafe CEO Diogo Almeida distrusts benchmarks and urges teams to test models on their own work, though TypeSafe itself released rankings alongside its latest Jev model, which adds image support and smart-if statements. Public indices—the AIM System One Index and the Artificial Analysis Decision Index—favor Liquid’s D1 Ultra and Base Lab’s Elephant-1; the latest Jev is seventh and third, while OpenAI‘s decision model is eighth and fourteenth. TypeSafe‘s offerings remain the most widely queried on OpenRouter. The most whimsical benchmark, JevMan, has decision models play PacMan, and the latest Jev tops that ranking. Felix Wunderlich, chief growth officer at Opper AI, which built the benchmark, said PacMan is an interesting test because of its nonstop stream of quick, high-stakes decisions: at every junction the model must decide where to go within a few hundred milliseconds.

OpenAI‘s potential edge is multi-modality. For image classification, Handa said, “Luna is extremely powerful and has been showing better calibrations,” and customers want the Decisions API to be more flexible about using images as outputs as well. He also sees further speed opening new use cases: “Can we get this to something like 60 milliseconds, and then you could imagine an API call for every single frame in a video?” As offerings get faster and cheaper, they may live up to Jev‘s namesake—an economist who argued that using a resource more efficiently makes it more ubiquitous. Handa expects an expanding “crazy set of use cases” that were hiding in plain sight until decision models revealed them.