Windows x64 native binary for @paniolo/cli (Vulkan GPU build; falls back to CPU when no GPU is present).

Windows x64 native binary for @paniolo/lasso (Vulkan GPU build; falls back to CPU when no GPU is present).

Instagram Stories-style video/image editor for Expo — color filters, text & sticker overlays, music mixing, and FFmpeg export, with native iOS color grading (no GPU readback).

In-process GGUF execution adapter for the Graphorin framework. Wraps node-llama-cpp@^3.5 to load .gguf model files directly into the same Node process - no daemon, no port to manage, no GPU contention with other processes. Declares trust: 'loopback' perma

Create thumbnail from 3D STL file. Creates beautifully rendered png and jpeg output server-side with no GPU from ASCII and Binary STL's.

Classify GPU's based on their benchmark score in order to provide an adaptive experience.

OHAM — huge pictures and video, usable on the device that receives them. A 423-megapixel photo is a 712 KB file that opens on a phone; 4K plays at 60 fps on a phone CPU with no GPU and no codec. The `oham` command line reads sealed .tsb containers entirel

Bela — small AI with big ambitions. A self-training, app-scoped agent engine: 1.5M parameters, trains in seconds on one CPU core, runs a real hotel. No GPU, no LLM API, MIT.

The luma.gl core Device API

VZT Video-Intel — temporal scene-graph MCP + CLI that gives Claude video understanding. Zero install. No Docker. No GPU. Lite mode (free, offline WASM) or cloud mode (Replicate).

ONNXRuntime Node.js binding

[](https://www.npmjs.com/package/stats-gl) [](https://www.npmjs.com/package/st

KTX 2.0 (.ktx2) parser and serializer.

Lightweight local AI coding agent CLI — an autonomous terminal agent running Qwen2.5-Coder 3B fully offline on CPU (Linux + Windows, no GPU). OpenAI-compatible, no API keys.

This repository provides native TensorFlow execution in backend JavaScript applications under the Node.js runtime, accelerated by the TensorFlow C binary under the hood. It provides the same API as [TensorFlow.js](https://js.tensorflow.org/api/latest/).

Train small, local text classifiers from labeled examples. No API keys, no cloud, no GPU.

Path tracing renderer and utilities for three.js built on top of three-mesh-bvh.

Classify GPU's based on their benchmark score in order to provide an adaptive experience.

A simple mocker for testing threaded development

GOCR — small, CPU OCR: document -> text + bounding boxes as JSON. ~30 MB, no GPU. Node port of the g-ocr Python package.