# Unsloth

> Apache-2.0 core for fast local LLM fine-tuning, plus an AGPL desktop studio

**Category:** Python  
**Pricing:** open-source  
**URL:** https://unsloth.ai/docs  
**GitHub:** https://github.com/unslothai/unsloth  
**Added:** 2026-10-18

---

## The Hook

Unsloth is a Python library and desktop app for fine-tuning and running open models on your own GPU, with memory-saving kernels that let LoRA and QLoRA training fit on consumer and single-node hardware.

## What's Good

Dual-licensed: the core unsloth package is Apache-2.0, optional parts such as the Unsloth Studio UI are AGPL-3.0 (unslothai/unsloth). Release v0.1.908-beta (2026-10-15), active since 2023 with near-daily releases. It supports NVIDIA, AMD and Intel GPUs, CPUs and Vulkan, on Windows, Linux, WSL and macOS. It trains and runs LLM, embedding, audio and diffusion models, exports to GGUF for llama.cpp and Ollama, and documents using local models with Claude Code, Codex and MCP. Free Colab notebooks cover most popular models. ★~78.2k.

## The Catch

Releases are tagged beta and move fast, so pin versions in training scripts. Fine-tuning still needs a capable GPU for anything beyond small models, and multi-GPU support is newer than the single-GPU path. Watch the license split: the library is Apache-2.0, but the Studio UI is AGPL-3.0. The one-line installer pipes a remote script into your shell, so read it first.

## Verdict

The default way to fine-tune an open model on one GPU — pin the version and know which part is AGPL.