Gpu GitHub Repositories (2026) · Open Source Tools

Java Inference Tornado toolkit: High performance LLM inference & serving for the JVM on GPUs

:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.

Distributed AI Model Training and LLM Fine-Tuning on Kubernetes

Molecular dynamics and Monte Carlo soft matter simulation on GPUs.

Tooling for optimized, validated, and reproducible GPU-accelerated AI runtime in Kubernetes

AI agent builder with a VM at its heart. Every agent gets its own microVM sandbox to install software, run code and use the GPU in. Python, Node.js and Rust.

Starling — a new Linux desktop environment: Swift shell, its own compositor, a Flutter-to-Swift framework port, and first-party apps

Kubernetes operator for self-hosted LLM inference across a heterogeneous GPU fleet: NVIDIA CUDA, AMD Vulkan, and Apple Silicon Metal. Runtimes: llama.cpp, vLLM, TGI, mlx-server. Multi-GPU sharding, model caching, OpenAI-compatible endpoints. Apache-2.0, run across homelab and on-prem fleets, actively developed.

Keep Python on your laptop. Run PyTorch operations and hold tensors on a remote GPU, including from a Mac with no CUDA installation.

KEDA External gRPC Scaler for GPU workloads - native NVML metrics via DaemonSet, no Prometheus required

A lightweight, private system monitor that lives in your macOS menu bar. CPU, memory, battery, network, disk, GPU, temperature, fans. Zero telemetry.

Find why PyTorch training is slow, compare runs, and catch performance regressions in CI.

A small, fast code editor written in Go, with a native GPU-drawn UI (no webview, no Electron)

Open-source agent runtime — SSH-native isolation, eBPF egress policy, Kubernetes + LXC backends, GPU passthrough, MCP-native CLI

Multi-platform high-performance compute language extension for Rust.

LUPINE is a GPU over IP bridge allowing GPUs on remote machines to be attached to CPU-only machines.

Decides who gets your GPU. Chat, image and voice engines lease the card in turn instead of grabbing it, so two of them never load at once and crash each other. One Go binary, nothing else to install.

Tensor Fusion is a state-of-the-art GPU virtualization and pooling solution designed to optimize GPU cluster utilization to its fullest potential.

Command-line utility for monitoring GPU hardware.

NVSentinel detects and remediates GPU faults on Kubernetes nodes