Get Free Community License
Local AI Agent Platform for .NET

Build AI Agents.
Ship with Confidence.

The complete SDK for building production AI applications in .NET. Agent orchestration, document intelligence, RAG pipelines, and 56 built-in tools in a single NuGet package. Zero cloud dependency.

Download Free View Documentation
100% On-Device
GPU Accelerated
Cross-Platform
DocumentQA.cs 
using LMKit;
using LMKit.Retrieval;

// Load your models
var chatModel = new LM("path/to/chat-model.gguf");
var embedModel = new LM("path/to/embed-model.gguf");

// Create document Q&A with RAG
using var docChat = new PdfChat(chatModel, embedModel);

// Load your documents
await docChat.LoadDocumentAsync("contracts.pdf");
await docChat.LoadDocumentAsync("reports.pdf");

// Ask questions with grounded answers
var result = await docChat.SubmitAsync(
    "What are the payment terms?");

Console.WriteLine(result.Response.Completion);
// Sources: contracts.pdf, p.12
CUDA Vulkan Metal Windows macOS Linux

A Complete AI Stack with No Moving Parts

LM-Kit.NET is a unique full-stack AI framework for .NET that unifies everything you need to build and deploy AI agents with zero cloud dependency and zero external dependencies. It combines the fastest .NET inference engine, production-ready trained models, agent orchestration, RAG pipelines, and MCP-compatible tool calling in a single in-process SDK for C# and VB.NET.

Built by experts, updated continuously. Our team ships the latest advances in generative AI, symbolic AI, and NLP research directly into the SDK. Not every problem requires a massive LLM. Dedicated task agents deliver faster execution, lower costs, and higher accuracy for specific workflows.

Category of one: LM-Kit.NET is the only .NET SDK that unifies a complete inference engine, production-ready trained models, agent orchestration, RAG pipelines, and MCP-compatible tool calling. Your AI. Your data. On your device.

56
Built-in Tools
18
Agent Templates
100+
Model Cards
0
External Dependencies

AI Agents and Orchestration

Build autonomous AI agents that reason, plan, and execute complex workflows within your applications. Complete agent infrastructure with Agent, AgentBuilder, AgentExecutor, and AgentRegistry for production-ready AI agents.

Agent Framework

Complete infrastructure with Agent, AgentBuilder, AgentExecutor, and AgentRegistry for building production-ready AI agents.

API Reference

Multi-Agent Orchestration

Coordinate multiple agents with PipelineOrchestrator, ParallelOrchestrator, RouterOrchestrator, and SupervisorOrchestrator.

API Reference

Planning Strategies

Multiple reasoning approaches: ReAct, Chain-of-Thought, Tree-of-Thought, Plan-and-Execute, and Reflection handlers.

API Reference

56 Built-in Tools

Ready-to-use tools: Data (JSON, XML, CSV), Text (Diff, Regex), Numeric (Calculator, Stats), Security (Hash, JWT), IO, Network.

API Reference

Agent-to-Agent Delegation

Enable agents to delegate tasks to specialized sub-agents with DelegationManager and DelegateTool.

API Reference

18 Agent Templates

Pre-built templates: Chat, Assistant, Code, Research, Analyst, Planner, Writer, Reviewer, and more for rapid development.

API Reference

Resilience Policies

Production-grade reliability with Retry, Circuit Breaker, Timeout, Rate Limit, Bulkhead, and Fallback policies.

API Reference

Streaming Support

Real-time response streaming with buffered, multicast, and delegate handlers for responsive UIs.

API Reference

Agent Observability

Full tracing and metrics with AgentTracer, AgentMetrics, and JSON export capabilities for debugging and monitoring.

API Reference

MCP Client Support

Connect to Model Context Protocol servers for extended capabilities including resources, prompts, and tool discovery.

API Reference

Agent Memory

Persistent memory that survives across conversation sessions with RAG-based recall for context-aware responses.

API Reference

Function Calling

Let models dynamically invoke your application's methods with structured parameters and type-safe contracts.

API Reference
Explore Agent Orchestration
Model Support

Run Any Model, Anywhere

100+ pre-configured model cards plus support for any GGUF model from Hugging Face.

Text Models

  • LLaMA 3.1 / 4
  • Mistral / Mixtral
  • Qwen 2.5 / 3
  • Phi 3 / 4
  • Gemma 2 / 3
  • Granite
  • DeepSeek R1
  • Falcon

Vision Models

  • Qwen-VL / Qwen2-VL
  • MiniCPM-V
  • Pixtral
  • Gemma Vision
  • LightOnOCR

Embedding Models

  • BGE-M3
  • Nomic Embed
  • Qwen Embedding
  • Gemma Embedding
  • Nomic Vision

Speech Models

  • Whisper Tiny
  • Whisper Base
  • Whisper Large V3
  • Whisper Large Turbo

Browse the Full Model Catalog

Explore production-ready models optimized for different tasks and hardware configurations. Load models directly from our catalog or any Hugging Face repository.

View Model Catalog Hugging Face
Why LM-Kit.NET

Complete Data Sovereignty by Design

Running inference locally provides inherent advantages that cloud solutions cannot match.

Complete Data Sovereignty

Sensitive information stays within your infrastructure. No data transmission, no third-party access, full audit trail.

Zero Network Latency

Responses as fast as your hardware allows. No round trips to cloud servers, no network dependencies.

No Per-Token Costs

Unlimited inference once deployed. Predictable costs regardless of usage volume or scale.

Offline Operation

Works without internet connectivity. Deploy in air-gapped and disconnected environments.

Regulatory Compliance

Meets GDPR, HIPAA, SOC 2, and data residency requirements by design. Audit-friendly operations.

Simple Deployment

Single NuGet package. No Python runtime, no containers, no external services to manage.

Core Differentiators

Document Intelligence & RAG Built In

Turn documents into knowledge. Ask questions. Get grounded answers with source attribution.

Document Intelligence

Complete document processing pipeline. Chat with PDFs, extract structured data from invoices and contracts, understand layouts with adaptive OCR/VLM processing.

  • PDF Chat and Document Q&A with retrieval, reranking, and grounded generation
  • Schema-based structured data extraction with JSON output
  • OCR and extraction pipelines for invoices, forms, IDs, emails
  • Native PDF, DOCX, XLSX, PPTX, HTML, and image support
  • Layout-aware processing with paragraph and line detection
Document Chat Document RAG Data Extraction

RAG & Knowledge

Ground AI responses in your organization's knowledge with a flexible, extensible RAG framework. Modular architecture for custom retrieval strategies.

  • Modular RAG architecture: use built-in pipelines or custom strategies
  • Built-in vector database without external dependencies
  • Multimodal RAG: retrieve from both text and images
  • Advanced chunking: Markdown-aware, semantic, layout-based
  • Reranking for precision, Qdrant integration for scale
RAG Pipeline Data Extraction Vector Database

Multimodal Intelligence

Process and understand content across text, images, documents, and audio. Build voice-driven assistants, visual inspection systems, and document understanding pipelines.

Vision Language Models (VLM)

Analyze images, extract information, answer questions about visual content with state-of-the-art VLMs.

API Reference

VLM-Based OCR

High-accuracy text extraction from images and scanned content using vision language models.

API Reference

Speech-to-Text

Transcribe audio with voice activity detection, multi-language support, and hallucination suppression.

API Reference

Document Processing

Native support for PDF, DOCX, XLSX, PPTX, HTML, and image formats with layout understanding.

API Reference

Image Embeddings

Generate semantic representations of images for similarity search and multimodal retrieval.

API Reference

Language Detection

Identify languages from text, images, or audio for automatic routing and processing.

API Reference
Explore Multimodal Capabilities

Content Intelligence & Text Analysis

Analyze and understand text and visual content. Compliance-focused text intelligence for PII extraction, NER, classification, and sentiment analysis.

Sentiment and Emotion Analysis

Detect emotional tone from text and images. Identify sentiment, emotions, and sarcasm.

API Reference

Custom Classification

Categorize text and images into your defined classes with zero-shot or fine-tuned classifiers.

API Reference

Keyword Extraction

Identify key terms and phrases from documents for indexing, tagging, and search optimization.

API Reference

Named Entity Recognition (NER)

Extract people, organizations, locations, dates, and custom entity types from text.

API Reference

PII Detection & Extraction

Identify and classify personal identifiers for privacy compliance: SSN, email, phone, addresses.

API Reference

Content Summarization

Condense long content with configurable strategies: extractive, abstractive, and hybrid approaches.

API Reference
Explore Content Analysis

Text Generation and Transformation

Generate and refine content with precise control. Build context-aware chatbots, constrain outputs with schemas, and transform text for your specific needs.

Conversational AI

Build context-aware chatbots with multi-turn memory, function calling, and smart context management.

API Reference

Constrained Generation

Guide model outputs using JSON schemas, templates, or custom grammar rules for structured outputs.

API Reference

Translation

Convert text between languages with confidence scoring and quality assessment.

API Reference

Text Enhancement

Improve clarity, fix grammar and spelling, adapt tone, and enhance writing quality.

API Reference
Explore Text Generation

Performance and Hardware

The fastest .NET inference engine. LM-Kit.NET automatically leverages the best available acceleration on any hardware with optimized kernels for maximum throughput.

NVIDIA GPUs (CUDA)

Optimized CUDA backends with custom kernels for maximum performance on NVIDIA hardware.

Apple Silicon (Metal)

Metal acceleration for M-series chips with unified memory architecture support.

Cross-Vendor GPUs (Vulkan)

Vulkan backend for AMD, Intel, and other GPU vendors with broad compatibility.

Multi-GPU Support

Distribute models across multiple GPUs for larger models and higher throughput.

CPU Fallback

Optimized CPU inference with SIMD acceleration when GPU is unavailable.

Dual Backend Architecture

Choose llama.cpp for broad compatibility or ONNX Runtime for optimized inference.

Platform Support

Run Anywhere .NET Runs

Full cross-platform support with hardware acceleration on Windows, macOS, and Linux.

Operating Systems

Windows 7 through 11, macOS 11+ (Intel and Apple Silicon), Linux with glibc 2.27+ (x64 and ARM64).

Windows macOS Linux ARM64

GPU Acceleration

CUDA for NVIDIA, Metal for Apple Silicon, Vulkan for cross-vendor GPU support.

CUDA Metal Vulkan

.NET Frameworks

.NET Framework 4.6.2 through latest .NET releases. Full MAUI support for cross-platform apps.

.NET 8-10 .NET Framework MAUI

Zero Dependencies. One NuGet Package.

The entire AI stack runs in-process within your .NET application. No Python runtime. No containers. No external services. No native libraries to manage separately.

$ dotnet add package LM-Kit.NET
NuGet Package GitHub Samples Semantic Kernel Data Connectors
Resources

Get Started in Minutes

Comprehensive documentation, runnable samples, and API reference.

Documentation

Complete guides covering agents, RAG, document processing, and deployment.

Read Docs

API Reference

Complete API documentation for all namespaces, classes, and methods.

View API

Code Samples

50+ runnable demos covering agents, chat, RAG, vision, speech, and more.

View Samples

Model Catalog

Browse pre-configured models optimized for different tasks and hardware.

Browse Models

Ready to Build Local AI Agents?

From prompts to production with zero cloud dependency. Download the SDK and start building in minutes.

Download Free Contact Sales