🖥️ 90+ private, self-hosted AI tools that replace the paid chatbot — offline, nothing leaves your machine

Tutorials & Methods
, ,
1

Private AI that lives in your own machine — offline, no subscription
Private AI that lives in your own machine — offline, no subscription1672×941 522 KB

:locked: Research desks that cite your files, agents that run your desktop, whole offline suites — plus the plumbing to route it all

Private AI on your own machine — offline, no subscription, nothing you type ever leaving the room. 90+ tools, sorted — the rarest first:

:spiral_shell: ANARCHY — a fully offline AI shell that runs your desktop: voice, RAG, Git, automation, permission tiers.
:microscope: Qube — research-grade private answers with real, inspectable citations, on its own engine.
:card_index_dividers: Odysseus — a whole private suite: chat, agents, email, notes, calendar — all local.
:honeybee: lilbee — an entire local AI stack (runtime + vector index + crawler) in one executable.
:locked_with_key: Akana — a self-hosted server where every memory write needs your approval, behind an encrypted vault.

:house: The stack — private AI on your own hardware

🧱 The engine underneath — pick one, the rest runs on it
  • Ollama — one command to pull + run a model; the backbone most tools sit on.
  • LM Studio — friendly desktop model lab plus a local API server.
  • Jan — offline ChatGPT-style desktop app, no terminal.
  • Open WebUI — the ChatGPT-style browser front-end for any of them.
  • llama.cpp — the raw engine underneath; runs on CPU when there’s no GPU worth using.
📚 Chat with your own files, privately — cited answers, no cloud
  • Qube — native desktop assistant, own GGUF engine, editable memory, inspectable citations.
  • lilbee — a whole local stack (runtime + vector index + crawler) in one executable.
  • Chaty — offline GGUF/MLX chat with RAG, voice, and deep-research.
  • GGUF Loader — offline models that act on a project folder, with cited RAG.
  • Beacon — Ollama + LanceDB + reranking, auditable answers from your files.
  • Khoj — a self-hostable “second brain” over your docs, custom agents, local backends.
  • AnythingLLM — self-hostable RAG workspaces + agents over local providers.
  • Gemma Genie — CLI Q&A over PDFs, Office files, images and audio, on-device.
🤖 AI that acts on your machine — agent desks, not chat boxes
  • ANARCHY — fully local shell: voice, RAG, Git, desktop automation, permission tiers.
  • Feral — hackable local agent desk with a sidecar runtime + deep-research.
  • nanobot — build a private assistant network: WebUI, tools, memory, MCP, connectors.
  • Local Agent Studio — wires local Ollama + ComfyUI with sandboxed tool execution.
  • Locally Uncensored — one installer: chat, coding, image/video, voice, RAG.
  • Pengy — one shell (GUI/CLI/web) routing several local model servers.
  • DARIOS AI — voice + local LLM + document memory + DevSecOps tools.
🗂️ A whole private workspace — suites that replace the SaaS, not just the chat
  • Odysseus — chat, agents, research, docs, email, notes, calendar — all local.
  • Artha — a local agent that makes DOCX/XLSX/PDF files; an offline office assistant.
  • Akana — self-hosted server with review-gated memory + an encrypted vault.
  • Aster — private workspace with bounded, persistent agents.
  • Off Grid AI — local studio: llama.cpp inference, OpenAI-compatible gateway, transcription, image gen.
🖧 Serve the whole household — one local box, every device
  • Local AI — portable Ollama workspace, RAM-aware, LAN-served to phones + laptops.
  • BlackCortex Lite — one Windows box serves the family over the LAN, per-user history, PDF chat.
  • rag_unplugged — Dockerized stack (Ollama, ChromaDB, SearXNG, Telegram/WhatsApp), every block swappable.
  • rag-chatbot — a Raspberry-Pi RAG bot for a household or small office.
🧑‍💻 Lean, air-gapped & creative — small binaries, companions, multi-model
  • Claudette — single-binary coding agent; tested offline mode blocks outbound calls (air-gap-safe).
  • Pern — offline dev workspace with an embedded Llama server.
  • Sullybase — lean Flask chat with live hardware monitoring.
  • PolyUI — compare several local models side by side, offline.
  • SweetrollLM · Aria — offline character/companion chat, no cloud accounts.
  • XandSuite — an all-in-one offline playground: GGUF, RAG, multimodal tools.
🌐 No machine to spare? No-install browser chats (no card, no signup)
🎯 Strong models, free in the browser — when you want frontier quality now
🔀 The plumbing — run, route & manage your local models

Your machine is the subscription now.

7 curtidas
2

:up_button: The Core-Community just AI-upgraded this topic.

1 curtida