Kev introduces tiny Jev-like decision models (0.8B, 4B, 9B) based on Qwen3.5, released under Apache-2.0 with training code and evaluation suites. The 9B variant runs in bf16 on a 32 GB Mac and is API‑compatible with TypeSafe’s System One, allowing the Python SDK to point to a local server unchanged. A web playground demonstrates how option ordering influences the model’s answers.
Read original
reddit/r/LocalLLaMA