Local & Self-Hosted AI Tools

RSI Harness builds on the Pi coding agent for developers and researchers who want to version and share tailored agent setups. It collects each setup into a local directory bundle called a Genome.

Numen provides hands-free voice control and dictation for Linux. It is for people who want to operate their computer through speech, including users who cannot comfortably type or want to reduce hand strain.

LMCache helps teams serving LLMs reuse context that a model has already processed. It reduces repeated work on long prompts, conversation history and retrieved documents, shortening the wait for the first token and improving throughput.

Anything Analyzer is a desktop protocol analyzer for developers and security researchers investigating APIs, authentication and request signing. It brings browser and app traffic into one session for inspection and AI analysis.

Hunyuan-A13B is Tencent's language model for developers and researchers building self-hosted AI applications. It handles general text and reasoning tasks, with optimization for AI agent workloads.

Link gives AI agents persistent memory so you don't have to repeat preferences, project decisions and context in every session. Multiple agents can read the same local Markdown files, which you can inspect in an editor or Obsidian.

JesType is a paid dictation app for people who want to write emails, documents, and other text by speaking while keeping their audio on their own computer. It runs on Apple Silicon Macs with macOS 13.4 or later and 64-bit Windows PCs running Windows 10 or later. Intel Macs aren't supported.

SpeakUp is a dictation app for people who want to write by voice on a Mac. It's suited to confidential notes, everyday writing and reducing typing because of RSI or accessibility needs.

Mirrai is an AI companion for people who want to revisit unfinished conversations through a persona based on someone's chat history. It simulates their communication style, with a farewell process that gives the experience an optional endpoint.

ChatTTS generates conversational speech for developers and researchers building spoken LLM assistants or dialogue experiments. The released model is restricted to academic use and cannot be used commercially.

Granite is IBM's family of AI models for developers and businesses building applications on their own hardware or servers. You can customize the models for enterprise tasks, including coding assistants and document analysis.

InvokeAI is a free, open-source image generation application for artists who want to create and edit artwork on their own hardware. Its locally hosted browser interface combines direct canvas work with repeatable generation workflows.

AIBrix provides infrastructure components for teams serving large language models on Kubernetes. It's aimed at organizations that need to manage inference capacity, request routing and models across a cluster.

Dograh is an open source voice AI platform for teams building agents that answer or place phone calls. It's a self-hosted alternative to Vapi and Retell, with control over conversation flows and a choice of AI providers.

Waku Agent is a personal AI assistant that runs on your laptop, stores its memory locally and calls remote model providers. It's for developers who want to inspect and change the assistant's behavior in Python. The agent uses provider API credentials for reasoning; requests to cloud providers are processed by those services.

Avibe gives coding agents a shared workspace for people who want to manage ongoing work and review results from their browser or phone. Agent execution stays on your machine, while inference uses your configured provider; OpenCode can also use local or OpenAI-compatible endpoints.

Xinference serves AI models on your own computer or servers for developers and researchers building applications around models they control. It brings chat, document search and media generation into one platform.

Voxta is an AI companion for people who want a character they can talk to, assign tasks to or use in interactive stories. You choose its personality, voice and optional avatar.

FastLLM is an open-source inference engine for people running large language models on their own machines or self-hosted servers. Its main draw is hardware flexibility: large mixture-of-experts (MoE) models can spread their weights across GPU memory, system RAM and disk when they don't fit entirely on a graphics card.

Kestra coordinates AI, data and infrastructure workflows for teams that need to connect tasks and track their execution. It can run existing scripts and containers, so workflows can span tools written in different languages.

StarDist detects individual cells and nuclei in crowded 2D microscopy images and 3D volumes. It's for researchers and bioimage analysts who need separate outlines for closely packed objects. Its shape-based approach suits objects that can be represented by star-convex polygons in 2D or polyhedra in 3D.

Kolega Code is a local-first AI coding assistant for developers working in the terminal or an editor. Its Gigacode engine lets a model write a Python workflow to coordinate sub-agents for repository reviews, migrations and tasks with independent workstreams. You can inspect and rerun those workflows, including parallel work and separate verification steps.

SwanLab is an open-source AI experiment tracking platform for researchers and model training teams. It brings training results into one place so you can spot problems and compare runs before choosing your next experiment.

AutoTrain supports model training on your own hardware through AutoTrain Advanced, a no-code application for people fine-tuning LLMs or building models from their data. Advanced can run locally or on cloud infrastructure. The project is no longer maintained, and it won't receive bug fixes or new features.

WritHer is a voice dictation app for people who want to write in Windows desktop apps and manage everyday tasks by speaking. It works offline after the initial model downloads, without an account.

ACDSee Photo Studio Ultimate is a paid Windows desktop photo editor for photographers who want to manage large collections and edit their photos in one application. It combines non-destructive RAW adjustments with layered editing and AI tools for finding related images, retouching portraits and generating content. It works directly with existing photo folders, without requiring a collection import.

AIdea for Zotero is an AI assistant for researchers who want to ask questions about papers and translate reading material inside Zotero. Its side panel works in the library view and the PDF or EPUB reader.

smart:limit is a proprietary AI audio limiter plug-in for mixing and mastering on macOS and Windows. It analyzes a track and suggests limiting settings, with loudness and dynamics meters in the same interface. It's aimed at producers and engineers preparing audio for release who want help controlling peaks while retaining detail in the mix.

Voice Changer (w-okada), also called VC Client, converts speech as you talk using AI voice conversion models. It suits people who want live voice changes, including gaming commentators running other demanding applications.

Unity MCP connects AI assistants to the Unity Editor so game developers can ask an assistant to make changes directly in a project. It works with local LLMs and MCP clients including Claude, Codex, Cursor and VS Code.

HippoRAG is a Python memory and retrieval framework for developers building LLM applications that need to connect facts across documents. It uses a knowledge graph to find related information for questions that require several pieces of evidence.

Whisperstream turns speech into text in the app you're writing in. It's for people who want to dictate messages, documents or coding prompts while keeping speech recognition on their computer.

Distil-Whisper is a family of speech recognition models for developers who want to transcribe English audio on their own hardware. The models build on Whisper, reducing model size and processing time with some differences in accuracy.

Devon is an open-source AI coding agent that runs on your machine and works across files in a project. It's aimed mainly at Python developers who want help making code changes and investigating bugs, with the ability to redirect the agent while it works.

MNN is a C++ framework for developers running AI models on phones, PCs and embedded devices. It supports inference and training on the device, with a small application footprint and hardware acceleration.

wllama lets developers add local LLM inference to browser apps without a separate inference server. It brings llama.cpp to JavaScript and TypeScript projects through WebAssembly, so models run on the user's device.

NeMo Guardrails is a Python toolkit for developers building AI assistants that need rules for what they say and which actions they take. It adds checks around conversations, document-based answers and tool use.

SAM 3D Objects reconstructs objects in 3D from a single image. It's for researchers and developers who want to recover object geometry and appearance from photographs, including cluttered scenes with partly hidden objects.

Transcribe Offline turns recordings and live speech into text for people who need to review conversations without uploading audio to a cloud service. It pairs transcript editing with audio playback in a desktop app.

Moondream is a vision language model for developers building apps that interpret images, such as warehouse or manufacturing footage. Its downloadable models can run on your own hardware, including offline in an air-gapped environment.

noScribe transcribes recorded interviews for qualitative researchers and journalists who need to keep audio on their own computer. It processes recordings locally rather than sending them to a cloud transcription service.

VoxelMorph is a Python library for researchers and developers who need to align medical images using learned deformations. It provides registration networks and training components, including unsupervised methods, with PyTorch and TensorFlow backends. The code is open source under the Apache 2.0 license.

Paritok is a self-hosted gateway for developers whose coding agents send large file reads, tool output and conversation histories to an LLM. It reduces that input before the request reaches the upstream model.

Audiblez converts EPUB e-books into spoken audio using Kokoro-82M on your computer. It's for readers who want an audiobook version of a book, with control over the voice and which sections get read aloud.

Sidekick is a native macOS app for people who want AI answers grounded in their documents and research materials. Its built-in llama.cpp engine runs GGUF models such as Qwen3.5 without a separate model runner.

DeePMD-kit is open-source software for molecular and materials researchers who want to train and run AI interatomic potentials on their own hardware. It learns atomic interactions from quantum-mechanical reference data, then uses those models in molecular dynamics simulations of molecules, solids and metals. It runs on CPUs and GPUs, with support for CUDA and ROCm source builds, and carries the LGPL-3.0-or-later license.