One of the creators of ChatGPT introduced a fundamentally new type of AI models. For two years, Diogo Almeida secretly developed Jev, a model that does not write text and does not generate tokens sequentially. It receives any unstructured data and simultaneously makes decisions specified by the developer. For example, selects one category from a list, assigns a rating, or determines the probability of a statement. Jev is also prohibited from using hidden chain of reasoning. Complex behavior must be assembled by the developer from many simple solutions using regular code. That is why the model works 20–200 times faster and costs 40–400 times less. A million input tokens cost 4.2 cents, and the output remains free forever (its calculation is so cheap that the company sees no point in counting it.) The company claims zero hallucinations. Jev cannot produce an answer outside the given options or break the structure. In this case, the model may choose the wrong option. This is not a replacement for GPT-6 Astra or a new chatbot. Jev is designed for automation, where programs do not need texts. They need quick solutions in a given format.