SHELFCRITTER AI Digest

MODEL

Kev

modeltopic-note

Overview

Kev is Jared Palmer’s open-source homage to TypeSafe AI‘s Jev, refactored onto Qwen3.5 with 0.8B / 4B / 9B checkpoints and no Jev outputs used in training. Landed on HN at 422 pts · 191 cmts (github.com/jaredpalmer/kev/tree/main) within 24 hours of Jev’s release. The corpus-relevant datum: an open-weights re-implementation of the new “decision model” shape shipped within hours of the closed release — the compression window for closed→open in this category is now measured in hours.

Timeline

  • 2026-09-22-AI-Digest — Kev landed on HN at 422 pts · 191 cmts — Jared Palmer’s open-source homage to TypeSafe AI‘s Jev, refactored onto Qwen3.5 with 0.8B / 4B / 9B checkpoints and no Jev outputs used in training. Pattern continuous with DeepSeek-style open-weights compression of closed frontier releases, extended into a category, not a version tier. Load-bearing test: whether Kev’s open-weights refactor gets pulled into downstream router / spam-filter / eval-router pipelines.

Key Developments

  1. 24-Hour Open-Weights Refactor of a Novel Model Shape: Kev shipped on Qwen3.5 within 24 hours of Jev‘s release, with an explicit “no Jev outputs used in training” claim — if it holds under scrutiny, another data point that the closed→open replication window for a novel model shape is now measured in hours, not weeks.

  2. Decision-Model Ecosystem Test: The load-bearing question is whether Kev’s 0.8B / 4B / 9B checkpoints get pulled into downstream router / spam-filter / eval-router pipelines — that adoption would elevate decision models from a one-vendor product to a genuine category.