▌system-one

~ / open-implementations / trained-replicas / kev

kev

LoRA plus a pointer readout on Qwen3.5 (0.8B / 4B / 9B), Apache 2.0, with training code and frozen eval suites. TypeSafe /v1/systemone drop-in, calibrated by default via a fitted temperature. Frozen evals put Kev-9B about 3.5 points behind Jev on unseen sources; MLX, ROCm, CUDA, and a one-command Modal deploy. Also the only open-weights route on OpenRouter's Decisions API.

section
Open Implementations
group
Trained replicas
source
github.com
repo
jaredpalmer/kev
stars
9,052 ▕███████████░▏
7-day gain
+697
forks
606
language
Python
license
Apache-2.0
last pushed
oct 12
added to list
sep 20

## history

2026-09-130 → 9.1k stars2026-10-13

also in Trained replicas