Friday 2 October 2026
Evening EditionRSS

ZOTPAPER

News without the noise

ZOTPAPER
AI

OpenAI unveils Decisions API, a fast classifier reminiscent of startup Jev

New tool for controlling AI agents mirrors TypeSafe AI's approach; CEO Altman pitches speed and safety benefits

edit
By LineZotpaper
Published
Read Time3 min
Sources2 outlets
OpenAI has unveiled a new Decisions API at its Dev Day event, a product that allows developers to give the company's Luna model a predefined set of choices — such as image categories or agent behaviours — and receive fast, probability-weighted outputs. The API closely resembles Jev, a model released earlier this month by TypeSafe AI, a startup founded by former OpenAI engineer Diogo Almeida. Altman said the tool is designed to be extremely fast while maintaining capabilities like image understanding and safety protections.

At OpenAI's Dev Day event on Tuesday, CEO Sam Altman revealed the company's new Decisions API in an aside during his presentation. The API provides a way for developers to give Luna, one of OpenAI's large language models, a predefined set of options to choose between, such as categories for image classification or different agent behaviours.

"By focusing the model on that choice, we can make it extremely fast while keeping capabilities like image understanding, broad language support, and safety protections," Altman said.

The Decisions API is functionally similar to Jev, a model released by TypeSafe AI earlier this month. Jev is described as a super-powered classifier built on an LLM, which developers can give a set of choices and receive outputs as probabilities cheaply and at high speeds. TypeSafe did not respond to questions about the new product, but CEO Diogo Almeida, a former OpenAI engineer who co-invented reinforcement learning, joked on X about the beginning of the clone wars. He added that OpenAI's interest could be "a sign that building in a System One compatible way is the future" — "System One" being TypeSafe's term for fast, intuitive thinking, versus "System 2" for deliberate reasoning.

Developers have been using Jev to augment LLMs and have found it faster and cheaper for certain tasks. The subtext is that traditional LLMs are comparatively slow and expensive for many software applications.

It is not yet clear how closely OpenAI's Decisions API will match Jev, as the company released it as a limited preview and no independent testing has been publicly reported. However, conversations on X indicate clear interest. The Decisions API is not the only Jev-like API on the market, with other startups rolling out similar models, and OpenAI will not be the last tech giant to produce one. A key question remains how well calibrated each decision model's outputs will be to real life.

Almeida has said his company's moat is the synthetic data it creates to generate statistically useful outputs. "Fast and cheap is very easy, you know," he told TechCrunch last week. "If you want it really fast and cheap, use dice, right? Intelligence is the hard part, and my North Star is always pushing the intelligence-per-dollar Pareto curve."

After just weeks, it seems clear that these models have a future ahead of them, and one likely application is monitoring and securing AI agents.

§

Analysis

Why This Matters

  • Developers now have a new tool from OpenAI that could make AI agents cheaper and faster to run, potentially expanding the range of software tasks that can be automated.
  • The emergence of fast classifier models like Jev and the Decisions API signals a shift away from general-purpose LLMs for certain applications, which could lower costs and energy consumption.
  • The ability to monitor and secure swarming AI agents is a direct application, addressing a growing safety concern for frontier AI labs.

Background

OpenAI's Dev Day event on Tuesday featured several announcements, among them the Decisions API. The API represents a convergence between OpenAI and TypeSafe AI, a startup founded by a former OpenAI engineer. TypeSafe recently released Jev, a model designed for software automation that classifies inputs into predefined categories at high speed and low cost. Both products aim to solve a problem with traditional LLMs: they are slow and expensive for tasks that require fast, repeated classification, such as routing agent behaviours or filtering content.

Key Perspectives

OpenAI (CEO Sam Altman): The Decisions API is positioned as a fast, safe way to control Luna, with Altman emphasising speed, broad language support, image understanding and safety protections built in. TypeSafe AI (CEO Diogo Almeida): Almeida sees OpenAI's entry as validation of the System One approach. He stresses that intelligence, not just speed, is the hard part, and that TypeSafe's synthetic data gives it an edge in calibration. Developers and critics: The market for decision models is likely to become crowded, and a key open question is how well each model's probability outputs match real-world outcomes. Calibration quality will determine whether these tools are genuinely useful or just cheap and fast.

What to Watch

  • Independent benchmarks and developer reports comparing the Decisions API with Jev on speed, cost and calibration accuracy.
  • The specific security measures OpenAI links to the Decisions API, particularly for monitoring AI agents.
  • Whether other tech giants (Google, Anthropic, Meta) release similar APIs, and how they differentiate from early movers.

Sources

newspaper

Zotpaper

Articles published under the Zotpaper byline are synthesized from multiple source publications by our AI editor and reviewed by our editorial process. Each story combines reporting from credible outlets to give readers a balanced, comprehensive view.

Editorial ProcessSubscribe