What it is
Kev is a family of small decision models (0.8B, 4B, 9B) built on Qwen3.5, with training code, evaluation data, and a web playground. It's for developers who want Jev-style typed decisions locally.
How it uses Jev
Kev is not Jev; it's a Jev-like replica. It exposes the same System One API: you send a state plus noul, choice, and score questions, and it returns probabilities, a chosen option, and a score. The TypeSafe Python SDK can point at a local Kev server.
Primitives:choicescorenoul
Technique worth stealing
Questions share the input text but can't read each other, so each decision is scored independently from the same state.
Try it
uv sync --extra serve, then run python -m kev.serve --run jaredpalmer/kev-4b --port 8009.
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.90
- Shows a System One patternyes0.67
- Handles uncertaintyno0.04
- Measuredno0.10
- Runnableyes0.80
- Worth recommendingno0.39
- Model replicayes0.97
- Problem scopescore on a 0–2 scale0.84
- About Jevyes0.96
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 21 Sept 2026.