Kev: Tiny Jev-like family of decision models built on top of Qwen3.5

Kev is a family of small decision-making models built on Qwen3.5 that can be trained and run locally, available in 0.8B, 4B, and 9B sizes with an API compatible with TypeSafe's System One. The models support multiple question types (yes/no, multiple-choice, and rating) in a single request and can run on both CUDA and Apple Silicon hardware. Users can access pretrained weights or train their own models, with a web playground available for testing different inputs and option ordering effects.