English
Español
Most of the attention on AI goes to the big, visible moments, the chatbot that writes your email or the agent that books your flight. But a lot of the real work is small and invisible: the thousands of tiny yes-or-no judgments that software makes behind the scenes every second. This week OpenAI made its Decisions API generally available, and it is built for exactly that, aimed squarely at a rival called Jev.
What the Decisions API does
Normally, using AI means asking a model a question and getting a full written answer back. That is powerful, but it is also slow and expensive if all you need is a quick call. The Decisions API strips that down. Running on a model called GPT-6 Luna, with native integration into GPT-Live, it is purpose-built for small, repeatable, low-latency decisions.
In practice that means things like checking whether a condition is met, choosing between a fixed set of options, or scoring a piece of text, an image, or a block of JSON against a rubric you define. Is this comment spam or not? Which of these three categories does this support ticket belong to? Does this photo contain a product? These are the unglamorous judgments that power moderation, routing, and scoring inside almost every modern app, and OpenAI wants each one to be nearly instant and nearly free.
Why OpenAI is going after Jev
OpenAI is not inventing this category. It announced the Decisions API at DevDay 2026 and made it generally available on October 23, and it is positioned directly as a response to TypeSafe’s Jev model, which has built a following among developers for this kind of structured, repeatable decision-making.
That competitive framing matters. When OpenAI moves into a space, it usually does so on scale and price, betting that it can make the same capability cheaper and more deeply integrated with the rest of its platform. If it succeeds, the cost of automating a single small decision drops toward zero, and the kinds of judgment calls that used to require either a human or a custom-built system become a simple API call. The open question is whether OpenAI’s version is genuinely cheaper and better than Jev, or just more convenient because it lives next to everything else OpenAI already offers.
What it means for your business
This is a builder’s story more than a consumer one, but the implications reach everyone. The repetitive judgment work inside a business, flagging, sorting, categorizing, scoring, is often exactly what junior staff and operations teams spend their days on. A cheap, fast decision engine makes it possible to automate that layer at scale.
If your company runs software that makes lots of small decisions, routing tickets, moderating content, approving routine requests, this kind of API is worth understanding. The upside is speed and cost: decisions that happen in milliseconds for a fraction of a cent. The caution is judgment: these systems are only as good as the rubric you give them, and handing over thousands of automatic decisions means being very clear about where a mistake is acceptable and where a human still needs to sign off.
The Decisions API will never make headlines the way a flashy chatbot does, but it may end up mattering more. It is the plumbing that lets AI quietly handle the small, constant stream of choices that keep software running. If OpenAI can undercut Jev on price and win developers over, the automation of everyday business decisions is about to get a lot cheaper, and a lot more common.
