Kaito Ito introduces laya-mlx, an open-source classification system similar to Jev, based on text output probabilities, 50x faster and running on-device with under 1G memory usage, ported to MLX.
laya-mlx is an open-source classification system similar to Jev, based on text output probabilities. It is ported to MLX and runs on-device with a maximum memory footprint of 1G.
The post shares 10 GitHub repos that wrap a decision layer around every session. Example: toolgate, a tool-call firewall where Jev answers 7 yes/no risk questions per call, mapping them to allow, ask, or deny. Link
A step-by-step guide to building your first Jev-based AI agent. It highlights that Jev's real power lies in letting the model make decisions, and shows how to expand on top of it.
The author reflects on their internet footprint and mentions developing an Ollama-based approach to host Jev-style models, sharing related practices and encounters.
Antoine Chaffin argues that paying so much achieves so little, with Luna far below Jev, an open decision model that can run locally or via a cheap API.
The open-source project open-annie enables GPT-Live conversations in the browser, using Jev to capture faces, gestures, and dances in about 120 ms, all rendered in real time. Code is publicly available. Link
This post shares Alvaro Cintas's video introduction to using jev-model-router within Claude Code, including a link to the distribution page (Claude Code Templates). Link
A resource collection about TypeSafe's typed judgment model Jev, covering getting-started examples, SDKs, agent tools, and independent evaluations to help developers understand and apply the model. Link
This guide shows how to make better decisions under uncertainty and build a Jev-style decision model from scratch, applicable to decision analysis, data science, and business strategy. Link