Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

jev-mini-bench

Mobile-first Hugging Face space for exploring tiny decision models: feel the latency gap, try a toy router and vote on repros.

Benchmarkprototype
View on GitHub ↗Live demo ↗

Overview

Jev Mini Bench is a static Hugging Face space described as a pocket explorer of the tiny decision-model wave that followed Jev, including open reproductions such as kev, tev, gev, lev and Clef. It lets visitors feel the latency gap the author cites (about 1200 ms for a big LLM versus about 12 ms for a tiny decision model), try an honestly labelled demo heuristic router in English and Spanish, and vote for a favourite community repro.

It is more of an explainer and community page than a rigorous benchmark, and it points to the jev-decision-index space for full benchmarks and to a 2026 CMU paper on JEV-as-a-Judge for the academic case for decision-only routing components.

What the decision model decides

TypeQuestionAnswers
Yes / noShould this support conversation be routed (single yes/no decision)?(inferred)yes · no

How it works

  1. Open the live space on a phone or desktop
  2. Compare simulated latency of a big LLM and a tiny decision model
  3. Try the demo heuristic router in English or Spanish
  4. Vote for a community decision-model repro

Strengths

  • Quick, visual introduction to decision models
  • Links to fuller benchmarks and research

Caveats

  • The router is a demo heuristic, not a real model
  • Latency figures are illustrative, not measured benchmarks

Getting started

Open https://iamdgarcia-jev-mini-bench.static.hf.space

At a glance

Type
Benchmark
Status
prototype
Language
JavaScript
Runtime
Static web page (Hugging Face Space)
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

Is this a real benchmark?

It is mostly an explainer; full benchmarks are linked at the jev-decision-index space.

Does the router call a model?

No. It is described as an honest demo heuristic router.

What repros does it cover?

Community decision models such as kev, tev, gev, lev and Clef.

Related projects

  • warped-sys1-lab

    Local SvelteKit workbench and CLI for organizing, editing, running and comparing Jev experiments defined as JSON.

    CLIbetaTypeScript
  • OpenJev (GPT-AGI)

    Jev-compatible open System One engine with a Claude Code-style terminal REPL showing live probabilities, plus an experiment harness.

    ProjectprototypePython
  • jev-behavior-study

    Independent field guide to Jev 1.13.0 behaviour across 11,621 requests: framing sensitivity, Snake and 3D city driving experiments.

    ProjectresearchPython, TypeScript
  • jev-report

    Independent 52-page Chinese report on Jev that checks vendor claims and reproduces 50 Chinese tests, with traceable data.

    ProjectresearchPython
  • jev-playground (Ahmar004)

    Interactive teaching site where Jev races frontier LLMs in games, an arena, a sandbox and quizzes, with a keyless replay mode.

    AppproductionTypeScript
  • typesafe-showcase★ 1

    Next.js site with six live Jev demos, each compared side by side with Gemini or Claude on latency, cost and agreement.

    DemobetaTypeScript