What is GenClass and where does it run?

GenClass is a small 32M-parameter typed-decision model and a Chrome extension built on it. You speak and your browser acts, often before you finish the sentence; everything runs locally, with the model executing in the browser via WebGPU/WASM and speech recognition on-device.

How does GenClass compare with Jev?

In the project's own same-inputs table GenClass wins at computer control (90.4% vs 66.4% recognising a command mid-sentence, 94.9% vs 75.5% picking the words to type) while Jev wins general held-out questions 94.7% vs 80.5%. GenClass reports zero option-order flips against Jev's 10-13%.

How does GenClass run in the browser?

The model is a ModernBERT/Ettin encoder with typed decision heads and per-option attention isolation; it answers choice, noul and score questions in one forward pass. The browser layer turns each partial transcript plus visible page elements into one request and asks before anything risky.

How do I install the GenClass Chrome extension?

Download genclass-0.1.0.zip from the v0.1.0 release, unzip it, open chrome://extensions, turn on Developer mode, click Load unpacked and pick the folder. The first run downloads the model (~57 MB) and Moonshine speech (~63 MB) once, and it starts in dry-run mode until you switch on Live.

Is GenClass affiliated with TypeSafe AI?

No. The README states GenClass is an independent open-source project, not affiliated with or endorsed by TypeSafe AI; "Jev" is TypeSafe's product name, used only to describe what GenClass is comparable to, and the model uses the same request format as Jev's System One API.

Who built GenClass and under what license?

The GitHub user MeharPro publishes it under the Apache-2.0 license; the repository was created 2026-10-04 and pushed the same day, with the model weights, Chrome extension zip and training pipelines included and a Chrome Web Store listing under review.

Jev directory

CommunityJev Tools & CLIs

GenClass

A 32M-parameter typed-decision model plus a Chrome extension that runs entirely on-device with WebGPU/WASM and Moonshine speech recognition. Its local server speaks POST /v1/systemone, so existing Jev clients work by changing the base URL; a benchmark table claims zero option-order flips.

See highlights Open github.com

Stars: 4
Language: Python
License: APACHE-2.0
Reading time: 3 min
Published by: Community
Author: MeharPro
Last verified: Oct 2026

MeharPro/GenClass

User repository on GitHub

View on GitHub

GenClass - The version of Jev that runs on your browser

Forks: 1
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity github chrome-extension local computer-use

Works forJev for Classification Jev for Realtime control Jev for Decision gating

Used inJev for Software & DevTools

Languages: Python JavaScript
Platforms: Chrome extension API Browser
Capabilities: Open sourceRuns locally

Highlights

Model executes in the browser with WebGPU/WASM; speech recognition uses Moonshine on-device by default.
Serves POST /v1/systemone with the same request and response shapes as Jev, plus choice, noul and score heads.
Benchmark table: GenClass 90.4% vs Jev 66.4% recognising commands mid-sentence; Jev leads general questions 94.7% vs 80.5%.
Per-option attention isolation is designed to stop option-order flips; the project reports zero flips against Jev's 10-13%.
Apache-2.0 and independent: the README states it is not affiliated with or endorsed by TypeSafe AI.

Quickstart

bash

git clone https://github.com/MeharPro/GenClass && cd GenClass
python3.12 -m venv .venv && .venv/bin/pip install -e ".[dev]"
.venv/bin/python -m jev_local.server.app --ckpt <weights dir>

Watch out

README reviewed 2026-10-17; the comparative benchmark is the project's own, the methodology and caveats live in BENCHMARKS.md outside this digest, the Chrome Web Store listing was still under review at review time, and a dry-run mode is on by default until you enable Live.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Jev driving real Chrome from Grok Bot

Hacker News: Cua asks which computer-use steps need a frontier model and which are fast decisions.

frabonacci95 pointsShow HN: CUA-S1 – A System One Model for Computer UseRead the thread on Hacker News

X: Voice-controlled computer use on a Mac

X: A computer assistant that listens

From the Jev guides

Original write-ups on what this entry is for.

Jev at home: voice pipelines, Assist routing, and sentinels

Oct 16, 2026

How Home Assistant conversation agents, macOS voice pipelines, and an advisory sentinel put a fast typed decision in the loop — one call per utterance, thresholds in code, and a daily token budget.

Jev Tools & CLIs#home-assistant#voice#automation

Read article

Search visibility, audited by decisions

Oct 16, 2026

Jev-based SEO tools split the crawl from the call: deterministic crawlers count and parse, while typed judgments rank fixes, place internal links, score search results, and track AI-answer mentions — with costs and caveats attached.

Jev Tools & CLIs#seo#search-visibility#audits

Read article

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.

Jev Tools & CLIs#gateway#vercel#cloudflare

Read article

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai

GenClass

Highlights

Quickstart

Questions about this entry

What is GenClass and where does it run?

How does GenClass compare with Jev?

How does GenClass run in the browser?

How do I install the GenClass Chrome extension?

Is GenClass affiliated with TypeSafe AI?

What does the Jev-compatible API server offer?

Who built GenClass and under what license?

Reactions & coverage

More Jev resources like this

From the Jev guides

From the Jev AI model community