Jev 1.13

Jev 1.13 is the current release of Jev, TypeSafe AI's flagship and the first of its System One models, named after the fast, intuitive thinking Daniel Kahneman described in Thinking, Fast and Slow. Instead of generating text for a person to read, it makes judgments for software to act on: you send a state (a string, a JSON object or an array of text) with a set of questions, and it returns typed answers with probabilities, not prose. Noul questions return the probability that a statement is true, Choice picks from options you define, and Score places the state on your own rubric, the last two with a full probability distribution and a confidence value. TypeSafe trains Jev with its own method, Reinforcement Learning for Calibrated Decisions (RLCD), so its probabilities are tuned against real outcomes. Every question is evaluated in parallel and in isolation against the same state, so adding questions barely changes latency. Jev is built for atomic questions composed in code: ticket classification and routing, intent detection, guardrail checks and rubric scoring in place of an LLM judge, with confidence deciding what runs automatically and what escalates to a person or a reasoning model. Input is text only and accuracy is strongest in English. A request holds 64K tokens in total, with at most 32K for the state plus the longest question.