Judgment Arena for Comparing AI Models

Shen Sean Chen built a judgment arena to compare Jev against Claude Opus, Haiku 4.5, and GPT-5.4 Mini on 15 human-labelled questions.

View on X

Shen Sean Chen@ShenSeanChen𝕏

𝗜 𝗿𝗮𝗰𝗲𝗱 𝗝𝗲𝘃 𝗮𝗴𝗮𝗶𝗻𝘀𝘁 𝗖𝗹𝗮𝘂𝗱𝗲 𝗢𝗽𝘂𝘀, 𝗛𝗮𝗶𝗸𝘂 𝟰.𝟱 𝗮𝗻𝗱 𝗚𝗣𝗧-𝟱.𝟰 𝗠𝗶𝗻𝗶, 𝗮𝗻𝗱 𝗯𝘂𝗶𝗹𝘁 𝘁𝗵𝗲 𝗷𝘂𝗱𝗴𝗺𝗲𝗻𝘁 𝗮𝗿𝗲𝗻𝗮 𝘁𝗼 𝗱𝗼 𝗶𝘁. Jev is @typesafeai's System One model, and it does something no chatbot does: it never talks to you. JSON in, JSON out, a probability on every answer. I drew System 1 and System 2 out on the whiteboard first, step by step, then put all four models on the same 15 human-labelled questions inside my ow

Sep 22, 2026X postsView on X

Jev is @typesafeai’s System One model, which returns JSON with a probability for every answer. Chen says he first mapped System 1 and System 2 on a whiteboard.

Also filed under Tools & apps

Labels LinkedIn Posts for AI Slop
Slop Radar labels LinkedIn posts as human, unclear, or AI slop as users scroll, with reasons shown on hover. It is a Chrome extension powered by Jev.
Hands-Free Recipe Page Control
Recipe Mode is a Chrome extension powered by Jev that lets users cook hands-free on any recipe page using commands such as “next,” “how much butter?” or “set a timer.”
Helm Release Risk Estimator
Jelm is a Jev-powered Helm release risk estimator for Kubernetes CI/CD. It scores deployment risk, analyzes rollouts, gates Gateway API canaries, and supports safer promote/rollback decisions.
Quilt Tools and JEV Experiment Lab
SuperInstance’s quilt-tools repository contains ten working Quilt tool prototypes and a springboard lab for JEV/MothQuantum experiments.