What is local-jev from ronitgupta138?

An offline drop-in alternative to the TypeSafe Jev System One decision models, written in Rust and run as a single binary. The README advertises zero API keys, zero cloud egress and zero rate limits, with decisions on CPU in under 18 ms.

How do I point an existing agent at it?

Set the base URL to the local server: export TYPESAFE_BASE_URL=http://localhost:8080/v1. The README says the REST API specification is identical, so clients can swap the base URL and run without cloud dependencies.

Which decision primitives does it support?

Choice, Score, Noul and Route — the primitives the README lists for fast agent decision models — served at endpoints such as POST /v1/systemone/choice and POST /v1/systemone/noul.

What does it claim about latency and cost?

Its comparison table lists under 18 ms single-pass CPU logits versus 70–180 ms cloud latency, and $0.00 versus $0.002 per call. These are the project's own README numbers, not an independent benchmark.

Does it really need no key or internet?

Per the README, yes: zero API keys, zero cloud egress and no rate limits beyond your hardware. That makes it suitable for privacy-sensitive or air-gapped workloads where diffs, logs and state should not leave the machine.

What is the project's maturity?

Early but released: crates.io lists v1.0.0, the README targets Rust 1.80+, and badges report 16/16 tests passing. It is MIT licensed, created and pushed on 2026-10-10, and the excerpt does not name the model behind the engine.

How is it different from the hosted Jev API?

The hosted service sends state to TypeSafe's cloud and bills per call; local-jev keeps everything on your CPU behind an identical REST surface, trading a cloud model for a local engine. The README does not document which weights it uses.

Jev directory

CommunityJev Repos & SDKs

Local Jev — ronitgupta138

A 100% offline, zero-cost Rust drop-in for the TypeSafe Jev System One API: one native binary serves Choice, Score, Noul and Route decisions on CPU in under 18 ms with no API keys, cloud egress or rate limits.

See highlights Open github.com

Stars: 2
Language: Rust
License: MIT
Reading time: 2 min
Published by: Community
Author: ronitgupta138
Last verified: Oct 2026

ronitgupta138/local-jev

User repository on GitHub

View on GitHub

⚡ 100% offline, zero-cost drop-in alternative to TypeSafe Jev System One decision models in Rust (<18ms on CPU)

Forks: 1
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

This is one of several Jev listings for the same project. The canonical entry is Local Jev — amithgc.

Tagscommunity catalog-import madewithjev rust offline local-inference

Works forJev for Classification Jev for Routing Jev for Scoring

Used inJev for Software & DevTools

Languages: Rust
Platforms: API
Capabilities: Open sourceSelf-hostedRuns locally

Highlights

Serves an identical REST API spec — POST /v1/systemone/choice and /noul among others — so you swap the base URL and existing agent code runs unchanged.
The README comparison claims under 18 ms single-pass CPU logits versus 70–180 ms cloud round trips, and $0.00 versus $0.002 per call.
Ships as one self-contained native Rust binary, Rust 1.80+, published on crates.io at v1.0.0 with a 16/16 tests-passing badge.
Aimed at air-gapped workloads: proprietary code diffs, internal logs and environment states never leave the machine.
Designed for agent stacks such as Claude Code, Hermes, MCP tools and Browser-Use by pointing their base URL at localhost:8080/v1.

Quickstart

bash

export TYPESAFE_BASE_URL="http://localhost:8080/v1"

Watch out

Copy reviewed against the truncated repo README on 2026-10-15; the excerpt cuts off inside the architecture blueprint. The latency, cost and privacy figures are the README's own comparison table, not an independent benchmark, and the README does not state which model or weights back the local engine.

Questions about this entry

Short answers drawn from the linked project or page.

From the Jev guides

Original write-ups on what this entry is for.

The Open Reimplementations: Local Models That Speak System One

Sep 22, 2026

Closed weights have not stopped the ecosystem from rebuilding the shape in the open. API-compatible servers, retrained heads on small models, and inference-time tricks that skip generation entirely.

Jev Repos & SDKs#open-models#research#local

Read article

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai