Run Laya in the browser
Laya is small enough to run in a browser tab with no server and no API key. The input never leaves the page. The trade-off is a download of several hundred megabytes on the first visit and much slower inference than a native runtime.
362 projects
testing-agent
g4rcez
Runs plain-language browser tests using agent-browser and local Laya or Jev decisions
agent-tester
g4rcez
Creates and runs YAML or JSON browser tests with agent-browser and Jev or Laya decisions
laya-guard-trial
Ayaannnnnnn
Lets users test Laya-based shell-command judgments for an agent safety hook
Freelancer-Protection-Client-Scanner
MuhammadAsad29
Scans Upwork job posts with Laya to identify risks and display safety indicators
laya-burn
lileeei
Runs and fine-tunes the Laya decision model natively in Rust using the Burn tensor framework
petsona-catalogue-ai
jeromtom
Classifies pet products using evidence-backed typed decision models, including Laya
home-router
cloudeeo
Routes home-assistant requests locally using Laya before running commands, search, or chat
NapQuest
adkbbx
Runs a 2D woodland sandbox game with local Laya AI
laya-weights
SemanticaBook
Hosts quantized ONNX weights for in-browser inference with the multilingual Laya model
laya-gomoku-k3
TechyLX
Runs Laya on a SpacemiT K3 board to power a browser-based Gomoku game
laya-localserver
smellship
Deploys the Laya decision model locally and connects it to UI automation workflows
laya-notes
itsdula
Tags local Markdown notes by running the Laya ONNX model on the user's computer
browser-agents
hiszpan400
Routes browser automation tasks to pluggable agents, including a Laya agent
next-draw
enabled528
Compares lottery-number candidates with a local Laya model in a deployable web application
spot-snake
Gitguy268
A browser-based snake game with an optional AI player powered by Laya through WebGPU
laya-service-desk-demo
JuanRetamales
Demonstrates a Laya model adapted to classify and triage Spanish technical-support incidents
layaForWeb
realizer-1966
Runs a quantized Laya decision model in the browser with ONNX Runtime Web
laya-workflows-ko
realizer-1966
Runs Korean workflow demos in the browser using the multilingual Laya decision model
laya-web-ko
realizer-1966
Runs a quantized Laya decision model in the browser with ONNX Runtime Web
laya-rag-judge-ko
realizer-1966
Runs browser-based RAG claim verification and evaluation with the Laya typed-decision model
layaAgent-ko
realizer-1966
Runs a browser-based agent using Laya for routine decisions and WebLLM when needed
layaForMemory-ko
realizer-1966
Uses Laya typed decisions in a browser-based layer that invalidates stale agent memories
laya-browser-grounding
Who-Code
A study and demo using Laya-MLX to ground browser automation actions on page elements
nothing-to-lose
p3bri4n
Simulates a zombie invasion where Laya drives zombie decisions and a local LLM guides survivors
Libraries
kevala is a zero-dependency Rust engine compiled to WebAssembly, with WebGPU kernels. It works from any page:
<script type="module">
import { Kevala } from "https://cdn.jsdelivr.net/npm/kevala@latest/js/src/index.js";
const kevala = await Kevala.load({ model: "laya", onProgress: console.log });
</script>
or pnpm add kevala. According to its README, the Laya int8 pack is 479 MB and is kept in browser storage after the first visit.
@r4ai/laya-web runs ONNX Runtime Web with WebGPU and a WebAssembly SIMD fallback, and is designed to run in a Web Worker:
npm install @r4ai/laya-web onnxruntime-web
Demos you can open
- layaForWeb: the English checkpoint as quantized ONNX (default build about 440 MB), with a live demo.
- open-jev-laya: multilingual Laya on Transformers.js (fp16 ONNX about 647 MB) with Gomoku, Big Two and a 3D maze.
- layaAsRagJudge: checks RAG claims against retrieved evidence entirely in the tab.
Speed and support
The layaForWeb README reports that a three-question call on the default WASM backend took about 2 to 5 seconds on a 2-core machine. In that project, WebGPU works only with the int4 build, because ONNX Runtime's WebGPU MatMulNBits kernel supports 2- and 4-bit weights. open-jev-laya tries WebGPU and falls back to WebAssembly.
Caveats
Quantized browser builds do not match PyTorch exactly. layaForWeb reports 97.9% top-answer agreement for all three of its quantized builds, with the largest probability gaps in int4. For a real-time loop, laya-pong keeps the model native. Its README gives the reason: the checkpoint wants about 2.4 GB resident in f32, and wasm32 has a 4 GB address space and no Metal.
More ways to use Laya