Jev: a new class of decision-making models instead of text

A new type of model called Jev has been released: it accepts text input but outputs only floating-point numbers — yes/no answers, scores and categories. It is extremely fast and cheap but cannot write a single sentence. The idea has already spawned alternatives Kev and Nimble, the latter now supported in Ollama.

• Jev takes text input but returns only numeric answers
• It classifies and scores instead of generating text
• Alternatives Kev and Nimble follow the same approach
• Nimble is now supported in Ollama and runs locally