wasm

jax-js is an open-source ML library and compiler for the web, bringing JAX-like numerical computing to JavaScript with WebGPU and WebAssembly kernels, all running client-side.

parakeet.wgsl enables fast, accurate NVIDIA Parakeet TDT 0.6B V2 speech transcription entirely in the browser using raw WebGPU compute shaders and SIMD WASM, with a live demo and open-source library.

The Firecrawl team open-sourced pdf-inspector, a fast Rust library that automatically determines whether a PDF is text-based or scanned, directly extracts text and converts it to Markdown, avoiding unnecessary OCR, supports table and multi-column layout detection, and provides Python/Node.js/Rust/WASM interfaces.

This article presents a tool that simulates TCP packet loss and congestion control in the browser using Go compiled to WebAssembly, enabling network resilience testing for web applications.

Weblings is a Rust compiler toolchain compiled to WebAssembly, enabling compilation and execution of Rust code directly in the browser via a web UI with a playground and Rustlings exercises.

ComfyUI-OpenCut is a new open-source plugin that integrates a full-featured video editor directly into ComfyUI, allowing multi-track editing, keyframes, effects, and real-time preview in the browser without server-side processing.

Wanix is a Wasm-native Unix sandboxing tool that lets you run and interact with real Wasm and x86 programs entirely in the browser using Web Components, inspired by Plan 9.

The author presents a pattern for sharing AI agents by compiling them into sealed WASM modules that only import a single inference function, allowing others to run the agent using their own local model without API keys or security risks.

The author details the challenges and solution for adding Go as a language to a browser-based code runner (dailyprog), explaining why the standard GOOS=js approach fails in a V8 isolate and how using GOOS=wasip1 with a minimal WASI host shim succeeds.

Google announces LiteRT.js, a high-performance JavaScript runtime for running AI models directly in the browser using WebAssembly and hardware acceleration, as an evolution from TensorFlow.js.

NoiseLang is a probabilistic programming language where every value is a distribution. It compiles to efficient Monte Carlo simulations using a JIT compiler and supports conditional inference.

Satteri is a Markdown processing pipeline that combines a fast Rust engine with flexible JavaScript plugins, available as WASM for in-browser use.

A browser-based Haskell-powered programmable CAD environment that runs via WebAssembly, requiring a large WASM blob download.

Ternlight is a 7 MB embedding model that runs in the browser via WASM, enabling fast semantic search without any API calls or servers. It ships as an npm package and can be used in three lines of code.

A web version of Rome Total War was created using Fable and Codex Rust, compiled to WASM and rendered with WebGPU, supporting ~20k units at decent FPS.

The author shares their experience learning Scheme and using Hoot to compile it to WebAssembly, encountering stability issues but successfully running physics simulations in the browser.

This article presents a detailed benchmark comparing ECS and OOP architectures in JavaScript for a 2D physics simulation, testing memory locality and performance across multiple dimensions including broad-phase algorithms and sorting strategies, with results on an M4 Mac.

LangChain has released a unified evaluation stack (Harbor + LangSmith) and an in-process code execution solution based on WASM+QuickJS, aimed at improving the evaluation reliability and execution safety of AI agents.

The article describes WATaBoy, a Game Boy emulator that uses just-in-time compilation to WebAssembly to achieve performance exceeding that of a native interpreter, serving as a proof-of-concept for JIT-to-Wasm in emulation.

A modular, high-performance Rust implementation of GraphRAG (Graph-based Retrieval Augmented Generation) with support for WebGPU acceleration and three deployment architectures: server-only, WASM-only (client-side), and hybrid.