TypeSafe AI Releases Jev: A System One Model That Returns Typed, Calibrated Decisions Instead of Text

TypeSafe AI has launched Jev, described as a “System One Model” aimed at returning typed, calibrated decisions rather than generating text. The company says Jev is transformer-based but not a large language model; users submit a system state and typed questions, and the API returns decisions with probabilities that code can branch on. The service is available as a hosted API in early access behind a waitlist, with no published weights or self-hosting option. TypeSafe claims Jev uses a new stack including parallel sampling and Reinforcement Learning for Calibrated Decisions (RLCD). It includes a confidence score from 0 to 1 and charges $42 per billion input tokens.