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

usejev

Runs the open LAYA decision model under Bun with native ONNX inference behind a TypeSafe-compatible API, plus an English/Persian playground.

Librarybeta
View on GitHub ↗

Overview

Laya on Bun runs the LAYA decision model locally using Bun, Hugging Face tokenizers and native ONNX inference. The HTTP server implements TypeSafe's decision API, so the official @typesafe-ai/sdk works by setting a custom baseURL. Like Jev, LAYA does not generate text; it answers choice, score and noul questions in one forward pass.

The monorepo includes a playground with 16 demos in English and Persian, such as routing support requests, classifying feedback and scoring urgency. Python is used once to export the pretrained model; serving runs entirely in Bun. The README warns of several GB of disk and RAM: about 820 MB of source weights and a 1.6 GB float32 ONNX model.

What the decision model decides

TypeQuestionAnswers
ChoiceRoute a message to billing or supportbilling · support
ScoreRate urgency from low to critical—
Yes / noIs there an explicit refund request?yes · no

How it works

  1. Install Bun and export the model once with Python.
  2. Start the local inference server.
  3. Point the official SDK at the local baseURL.
  4. Use typed answers; try the playground demos.

Strengths

  • Local inference with no API key.
  • Works with the official TypeSafe SDK.
  • Bilingual playground with 16 demos.

Caveats

  • Several GB of disk and RAM needed.
  • LAYA's accuracy and calibration may differ from Jev.

Getting started

bun install --frozen-lockfile, set up the export venv, then bun run model:download && bun run model:export && bun run smoke

At a glance

Type
Library
Status
beta
Language
TypeScript
Runtime
Bun server with ONNX Runtime
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

  • Run Laya locally with Bun: native ONNX inference, a TypeSafe-compatible API, and a bilingual decision playground.

    logicrw/awesome-jev-projects

  • Run Laya locally with Bun: native ONNX inference, a TypeSafe-compatible API, and a bilingual decision playground.

    hellogumbo/awesome-jev

  • Runs Laya under Bun with native ONNX inference behind a TypeSafe-compatible API, so the official SDK reaches it with a `baseURL`. Ships an English and Persian playground.

    MrJev/awesome-jev

Frequently asked questions

Do I need a TypeSafe key?

No, LAYA runs locally.

Can I use the official SDK?

Yes, set its baseURL to the local server.

How much space does it need?

Several GB; the ONNX model alone is about 1.6 GB.

Related projects

  • gojev

    Go harness for System One models: hosted Jev, open Kev in-process via llama.cpp, and Cloudflare Clef, with typed helpers and a CLI.

    ServicebetaGo
  • jev

    TypeScript library for one-line local decisions with open models (kev-0.6b, kev-4b, open-jev), no API keys or cloud.

    ProjectbetaTypeScript
  • jev-foundation-models

    Swift 6 bridge that runs System One decision models behind Apple's Foundation Models API, on device with Laya or via hosted endpoints.

    ProjectbetaSwift
  • laya-system-one

    Zero-dependency npm package that runs Laya natively or as WebAssembly in Node, Bun and browsers, drop-in for Jev's /v1/systemone.

    ProjectbetaJavaScript
  • OmniStudio

    Local-first desktop AI workbench for llama.cpp, vLLM and SGLang that includes a JEV typed-decision page and /v1/systemone gateway.

    Servicebeta
  • ollaya★ 1,305

    Rust daemon and CLI that pulls open decision models like Laya and serves them locally behind a TypeSafe-compatible /v1/systemone.

    ServicebetaRust