Free Open-Source AI Tools You Can Run Yourself (2026)
These are the tools you can run yourself. Self-hostable or local-first, with source you can read — which makes them the honest answer whenever the data you are working with is not yours to upload.
31 tools · free tiers last re-read July 2026 · how we verify
Free open-source LLM built for mathematical reasoning, scientific analysis, and accurate code generation.
Devstral2 provides 1.3B/123B parameter coding models with CLI agent for multi-file edits, Git integration, and 256K context.
Stable Diffusion 3.5 is the leading free open-source AI image generator supporting local installation and customizable templates for unlimited creativity.
NVIDIA research model that generates textured, production-ready 3D assets with PBR materials from text or image inputs in around two minutes.
Meta's 70-billion parameter open-source code generation model for multi-language coding, completion, and infilling.
Open-source multilingual LLM for reasoning, code generation, translation, and agentic workflow integration.
Generates full-minute videos from text storyboards with perfect temporal coherence.
Open-source AI world model by Decart and Etched that generates real-time Minecraft-style interactive gameplay at 20 FPS using next-frame prediction, with no traditional game engine required.
Stanford academic research model for high-resolution video generation using a shared image-video transformer architecture.
Google research model for video generation and editing using space-time diffusion for realistic motion synthesis.
YuE AI transforms lyrics into complete 5-minute songs with synchronized vocals/instruments in English, Chinese, Japanese, Korean.
StoryDiffusion GitHub AI generates coherent long-sequence images/videos maintaining character consistency, plot continuity using auto-attention.
Nvidia open-source AI research model that generates textured 3D shapes from 2D image collections without 3D supervision.
Free open-source AI from Meta FAIR that animates hand-drawn character sketches in the browser without animation software.
Gemma 3 is Google's open-weight multimodal AI model family supporting text, images, and video across 140+ languages under Apache 2.0.
Open-source Python framework with 52k+ GitHub stars for building autonomous AI agents with multi-agent orchestration, browser automation, MCP support, and custom tool integration.
Skywork AI's 1.8B open-source interactive world model generating real-time 25 FPS gameplay from keyboard and mouse inputs, with long-sequence consistency and free weights on GitHub and Hugging Face.
Open-source Python library that lets AI agents browse, click, and automate real websites programmatically.
Open-source ERP and CRM platform with integrated AI assistants covering sales, finance, inventory, HR, and 40,000+ business applications.
ByteDance AI research model that animates still human photos into realistic motion video while preserving facial identity and temporal consistency.
Meta AI generates talking avatars from text/audio with emotion control and multi-character conversations.
Combines multiple AI models and styles creating optimized hybrid illustrations from text.
Nvidia's open-source speech recognition model with automatic punctuation, precise word timestamps, and developer-friendly deployment.
Microsoft Research's open-source generalist multi-agent system with an Orchestrator directing four specialized sub-agents for web navigation, file handling, coding, and terminal execution.
Tencent's open-source AI model that generates interactive, action-controllable game video sequences from a single image and keyboard inputs.
Open-source CVPR 2025 AI model from Sony AI and UIUC that generates frame-synchronized audio from video and text inputs.
Open-source agentic AI coding model from Alibaba's Qwen team, available on Hugging Face for code generation and autonomous programming tasks.
Tencent's 13B model excels at 256K context reasoning, multilingual agents, and search.
Meta's powerful AI generates high-quality videos with synchronized audio from text prompts.
13-billion-parameter model generating cinema-quality videos with precise motion control, transitions, and special effects.
FREE code editor - GitHub Copilot AI/IntelliSense/Git/debugger. 70%+ developers. 100+ languages/extensions.
Why this list exists
Every hosted AI tool is a decision to send someone else your input. For client work, health data, unreleased material or anything under an NDA, that decision is often already made for you. Running the model locally removes the question entirely, and these tools let you.
The open source label is applied from each tool's own published terms, read by an editor rather than scraped. It is only added where the wording is explicit, so a tool missing from this page has not been ruled out — it means we did not find a clear statement either way, and we would rather leave it off than guess.
Common questions
Does open source here mean free?
It means the source is available and the tool can be run yourself. Most are free in both senses, but some open-source projects also sell a hosted version — the listing says which, and this page is drawn only from tools we classify as free.
Do I need a powerful computer?
It varies enormously and we do not pretend otherwise. Some run comfortably on a laptop; some want a dedicated GPU. Each listing describes the tool honestly rather than implying a uniform hardware floor.
Is my data private if I run these locally?
If the tool runs entirely on your machine, your input does not leave it. That is the point of the collection. Check each project's own documentation before trusting it with sensitive material — we link straight to the source.
More verified free collections
Or browse every free AI tool we have verified.