TensorStack

A modular .NET SDK for building AI-powered applications.

TensorStack provides reusable components for local AI inference, text generation, image generation, media processing, hardware acceleration, and application development through a unified developer experience.

Built for performance and flexibility, TensorStack powers all applications developed by TensorStack-AI, including Amuse, the official reference application and UI demonstration of the platform.

Features

Native .NET SDK
Local AI inference
Text generation and LLM integration
Stable Diffusion image generation
Image upscaling and extraction pipelines
Image, video, and audio processing
Hardware acceleration support
Python interoperability
Extensible provider architecture
Cross-platform development

Packages

AI & Inference

Package	Description	Docs
TensorStack.Common	Shared runtime and utilities	README
TensorStack.TextGeneration	Text generation and LLM integrations	README
TensorStack.StableDiffusion	Stable Diffusion image generation	README
TensorStack.Upscaler	AI image upscaling	README
TensorStack.Extractors	Feature extraction and analysis	README
TensorStack.Python	Python interoperability	README

Inference Providers

Package	Description	Docs
TensorStack.Providers.CPU	CPU execution provider	README
TensorStack.Providers.CUDA	NVIDIA CUDA execution provider	README
TensorStack.Providers.DML	DirectML execution provider	README

Media Processing

Image

Package	Description	Docs
TensorStack.Image	Shared image utilities	README
TensorStack.Image.Bitmap	Windows Forms `Bitmap` ↔ Tensor conversion	README
TensorStack.Image.BitmapImage	WPF `BitmapImage` ↔ Tensor conversion	README
TensorStack.Image.ImageSharp	ImageSharp `Image` ↔ Tensor conversion	README