A faster route to AI decisions
OpenAI has made its Decisions API available to all customers, extending access to AI models designed to make rapid choices among options. The offering enters a growing field that includes TypeSafe's Jev, as developers explore decision-focused models for tasks where speed and cost matter.
Rather than introducing a separately trained model, OpenAI adapted its ChatGPT Luna model to reduce processing time and inference costs. The company says the API can return results in roughly 100 to 180 milliseconds, compared with one to three seconds for some earlier workflows. The public release also includes voice capabilities.
Early users including Box, Datadog, Ramp and Speak have tested the system. Speak, which develops AI language-learning tools, reported cutting the time between actions from 1.4 seconds to about 0.13 seconds.
Performance priorities can vary by application, making real-world testing important for teams choosing among decision models. As these systems become quicker and more affordable, they could enable more responsive software and bring AI-powered decisions into a wider range of everyday workflows.
