OpenAI’s Jev clone could help the frontier lab stop its swarming agents
By Pradhyuman,
At OpenAI's Dev Day event, CEO Sam Altman announced the new Decisions API. The API gives OpenAI's Luna model a set of predefined options, such as image classification categories or agent behaviors. Altman stated that narrowed choices keep the model fast while maintaining safety protections and image understanding.
The tool works similarly to Jev, a model released by TypeSafe AI. TypeSafe AI CEO Diogo Almeida built Jev as a fast classifier for software automation. Almeida noted that standard large language models are slow and expensive for basic software choices.
Developers use these decision models to monitor AI agents. Shapor Naghibzadeh, who leads the startup QueryStory, built a demo that uses Jev to check agent actions against assigned tasks. Naghibzadeh reported that monitoring agent actions cost $2.94 with Jev, compared to $372 with a frontier model.
Maintained by Pradhyuman.
Filed under: OpenAI, Models & Research, Products & Agents