Gemma4 GitHub Repositories (2026) · Open Source Tools

Gemma 4 26B-A4B inference in ~2 GB of RAM on any M-series MacBook

🚀 Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support

A native .NET LLM inference engine and agent runtime for GGUF models. TensorSharp provides a console application, a web-based chatbot interface, iPhone App, and Ollama/OpenAI-compatible HTTP APIs for programmatic access. It supports Windows/MacOS/iOS/Linux with full GPU capability

OpenClaw alternative in your pocket

Real-time two-way voice translator for VRChat — LLM-based, VR subtitle overlay & OSC

Run frontier LLMs and VLMs locally on Qualcomm devices across NPU, GPU, and CPU with a few lines of code

Real-time multimodal AI pipelines on Apple Silicon

Agentic ✧ Gemma Inference for Android System Intelligence