Open-source tools for Models & training

Model inference, serving, fine-tuning and local model tools.

Check model licence terms separately from the code licence, then compare memory requirements and supported hardware with your setup.

59 projects

academiasd_loralab-trainerstudio academiasd · MIT
A one-install, one-launcher Flask app that lets you train your own LoRA AI image/video models on a budget consumer GPU, across nearly every major open image/video model.
agent-smith chengjun023 · MIT
A local router that sends easy coding tasks to a cheaper model and hard ones to a stronger model automatically, plus a native macOS floating window showing live Codex…
ai-engineering-course amitshekhariitbhu · Apache-2.0
A free, complete course that teaches AI engineering step by step, from machine learning basics through building and evaluating LLM-based agents.
ai-engineering-lab zorost · MIT
Zorost Intelligence AI Lab open-sourced a free, self-paced 24-week AI engineering course that takes a beginner from Python to production LLM/RAG/agent systems, with 43…
ai-system-design amitshekhariitbhu · Apache-2.0
A free, complete guide that teaches how to design real AI systems, step by step, from LLM inference and RAG to AI agents, guardrails, and cost optimization.
ailearningguide fazmin · MIT
A desktop app that teaches how modern AI works through 51 hands-on labs covering tokens, transformers, RAG, agents and fine-tuning, with plain-language lessons next to…
airllm lyogavin · Apache-2.0
Lets you run massive open-source LLMs, including ones with hundreds of billions of parameters, on a single consumer GPU with as little as 4-12GB of VRAM, no…
audio-studio-mcp kanaworksai-hue · MIT
An MCP server that lets Claude Code or any MCP client generate royalty-free background music and sound effects locally, using the Stable Audio 3 model on your own Mac…
auto-ontology-mcp gogoymh · Apache-2.0
Share receipts, notes, papers or screenshots with your AI assistant and it extracts the facts into a searchable personal knowledge base (originals kept in your own…
backtalk jaredrhod · AGPL-3.0
backtalk lets you talk to your Claude Code agent out loud instead of typing, using free local speech models for both listening and responding.
bla qybaihe · MIT
BLA protects shipped apps (Electron, JS, Python, WASM, native binaries) from AI reverse-engineering agents by making the analysis burn tokens and follow decoys instead…
colibri justvugg · Apache-2.0
A tiny pure-C inference engine that runs frontier Mixture-of-Experts models, from 744 billion up to 2.8 trillion parameters, on consumer hardware by streaming experts…
comfyui comfy-org · GPL-3.0
ComfyUI is a free, node-based visual interface for running diffusion models like Stable Diffusion locally, so you can build AI image and video generation workflows by…
daidocs kerneta · Apache-2.0
Gives your AI assistant a memory that lives as plain .dai files on your own disk, so you can open, grep, and keep it instead of trusting a black-box memory feature.
ds4 antirez · MIT
An open-source inference engine that lets you run DeepSeek 4 and other leading open-weight LLMs directly on your own Mac or PC, no cloud API or subscription needed.
dsh-plugin-subscriptions v1ki · MIT
This plugin lets you run DeepSeek Harness agents using the ChatGPT, Claude, or Grok subscription you already pay for, logging in via OAuth instead of buying separate…
freellmapi tashfeenahmed · MIT
FreeLLMAPI puts the free tiers of dozens of LLM providers behind one OpenAI-compatible API, automatically routing each request to an available model and falling over…
freetoken flashml-org · Apache-2.0
FreeToken lets you run massive, datacenter-scale open-weight AI models directly on your own gaming PC or laptop, with a desktop app and CLI that manage GPU/CPU memory…
heretic p-e-w · AGPL-3.0
Heretic automatically strips the built-in refusals out of open-weight AI models, co-minimizing refusals and quality loss, with no fine-tuning or knowledge of model…
intent-router angel291592 · MIT
Intent-Router makes AI agents ask a clarifying question (or pause) before acting on a vague request, instead of guessing.
kaggle-tpu-lab arahim3 · MIT
A recipe for serving the 27B-parameter Qwen3.8 model on Kaggle free TPUs with a public OpenAI-compatible endpoint and a 262k-token context window, no paid GPU.
kimi-k3-in-c fareedkhan-dev · Apache-2.0
Lets a 2.78-trillion-parameter AI model run inference on a regular laptop with no GPU and just 8GB of RAM, by rewriting the whole inference engine in dependency-free C.
labllm greninja9257 · MIT
LabLLM is a free native macOS app that lets you design a GPT-style model, train it from scratch on your own data using Apple Silicon, then fine-tune and chat with the…
llama-modchip synchopate · Apache-2.0
Lets you run a huge 284-billion-parameter AI model on an ordinary 64GB computer using only the CPU, no GPU required.
llm-inference-engineering amitshekhariitbhu · Apache-2.0
A growing free course by Amit Shekhar (founder of Outcome School) that breaks down how LLM inference actually works under the hood, from the basic token-generation loop…
llm-master youngyangyang04 · MIT
A free, structured roadmap of 150+ tutorials taking developers from their first LLM API call through building production-ready RAG pipelines and AI agents.
llmfit alexsjones · MIT
One command tells you which local AI models will actually run well on your specific computer, out of hundreds of options.
locoagent skyltliu · MIT
A lightweight, from-scratch coding agent with tool calling, persistent sessions, workspace awareness, memory, and support for multiple LLM providers.
ltx-2 lightricks · Apache-2.0
Inference and fine-tuning tools for an AI model that generates video with synchronized audio from a single package.
magnitude magnitudedev · Apache-2.0
Magnitude profiles your computer's hardware, recommends the best local AI models it can run, then downloads, tunes and runs them so your existing coding agent works…
mesh-avatar-studio shinshin86 · MIT
Give it one character illustration and a coding agent builds an animated talking avatar you can fine-tune in a local editor.
minimind jingyaogong · Apache-2.0
A free open-source walkthrough that trains your own ~64M-parameter language model from scratch on a single machine in about two hours.
model-optimizer nvidia · Apache-2.0
NVIDIA Model Optimizer is an open-source library that compresses large AI models, through techniques like quantization and pruning, so they run several times faster and…
moss-image openmoss · Apache-2.0
An open-source text-to-image AI model from the OpenMOSS research group that can place text and visual elements exactly where you want and render that text clearly and…
ms-cookbook modelscope · Apache-2.0
A free, open-source cookbook that walks developers from picking an open-source model through inference, fine-tuning, evaluation, RAG, and agents to shipping a real AI…
nari-qwen3-tts nari-labs · Apache-2.0
Nari Labs open-sourced a high-performance serving engine for Qwen3-TTS that starts speaking in under 50 milliseconds and handles 10 simultaneous voice-generation…
oido lokutor-ai · GPL-3.0
A free, open-source speech recognizer that runs entirely on a $5 microchip instead of the cloud, transcribing English and Spanish speech in real time with…
omlx jundot · Apache-2.0
Run and manage local language models on Apple Silicon from a macOS menu bar app.
opencode anomalyco · MIT
OpenCode is a free, open-source AI coding agent you run from your terminal, an alternative to tools like Claude Code or Cursor that works with any LLM provider.
opengrok onlyterp · MIT
opengrok lets you swap which AI model actually answers inside the Grok Bot app (Claude, Gemini, GLM, Qwen and more) mid-conversation, with every switch verified…
pi earendil-works · MIT
An open-source toolkit for building AI agents: one unified API across OpenAI, Anthropic and Google models, plus a ready-made interactive coding agent CLI.
project-maya mw00 · MIT
Lets anyone run a very large open-weight AI model privately on their own GPUs instead of relying on a cloud provider.
reef human-agent-society · Apache-2.0
Reef lets AI agents keep getting better simply from how you use them, by capturing interactions and feedback and turning them into versioned updates to the agent's…
remiqora inikolax · MIT
It runs AI music generation entirely on your own machine, turning a text prompt into a track and letting you split stems and edit in a built-in multitrack studio.
slotstream carloslfu · MIT
Runs a 125-billion-parameter language model on a Mac with far less RAM than it needs by streaming parts of the model from the SSD.
smolmaxxing hyperionaidev · MIT
smolmaxxing runs the biggest LLM it can on the smallest hardware you own, with measured benchmarks on a TV box, an old iPad, and a phone running 16B-80B parameter…
soup makazhanalpamys · Apache-2.0
Soup lets you fine-tune an 8-billion-parameter AI model on a laptop with just 4GB of GPU memory, using one YAML config file instead of wrestling with cloud setup.
stable-diffusion.cpp leejet · MIT
Lets you run popular image and video generation models like Stable Diffusion, Flux, and Qwen Image without Python or a GPU-heavy setup.
taskdistill b0yko · Apache-2.0
A tool that watches how you actually use an expensive AI model on one task, then trains a small local model to handle most of those calls itself, falling back to the…
time-to-first-token patchy631 · Apache-2.0
A free 10-week roadmap for learning how LLM inference serving actually works under the hood.
tokenspeed lightseekorg · MIT
A high-speed engine for running large language model inference.
transformers huggingface · Apache-2.0
The open-source library that provides ready-to-use implementations of virtually every major AI model for text, vision, audio, and multimodal use, letting developers run…
trnrt runanywhereai · Apache-2.0
An open-source engine that runs large language models efficiently on AWS Trainium chips, with an OpenAI-compatible API and support for 1-million-token context windows.
turbovec ryancodrai · MIT
A fast, compressed vector index (built on TurboQuant) for searching embeddings at scale, the kind of component that powers RAG and semantic-search systems.
uncut balanov · MIT
A tiny 7MB tool that lets you run AI coding agents from the command line or CI without losing any of their output to silent truncation or summarization.
unsloth unslothai · Apache-2.0
A free local app for running and training your own AI language and image models on your own machine.
vllm vllm-project · Apache-2.0
vLLM is an open-source engine that runs large language models fast and cheaply in production, used widely to self-host models like Llama, Qwen and DeepSeek.
voicebox jamiepine · MIT
A local-first, open-source AI voice studio that lets anyone clone a voice, generate speech, and dictate into any app on their own machine.
voicestudio debpalash · AGPL-3.0
A free, fully-local alternative to ElevenLabs that clones voices, dubs video, and creates audiobooks and transcripts across 646 languages, all on your own machine.