# Unsloth

Local UI to run and train LLMs and diffusion models, including Qwen3.8, Kimi K3, MiniMax-H3, Gemma 4, DeepSeek-V4, FLUX and more.

Repository: https://github.com/unslothai/unsloth
Canonical: https://ross.abutalabs.com/products/unsloth
Homepage: https://unsloth.ai/docs
Language: Python
License: Apache-2.0
License Family: permissive
Topics: fine-tuning, llama, llms, gemma, unsloth, llm, deepseek, text-to-speech, tts, qwen, agent, openai, reinforcement-learning, self-hosted, ui, image-generation, stable-diffusion, chatgpt, ai, python
Last push: 2026-08-27T00:28:18+00:00

## Health v2 (maintenance only)
Score: 94/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 72
- inputs: {"age_days": 1008, "days_push": 7, "days_rel": 8, "gap_med": 6.5, "n_releases_24m": 53}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 74883, forks 6781 (observed 2026-08-28T04:12:21.483761+00:00)

## What it is
Unsloth is a desktop application for running and fine-tuning LLMs, diffusion, embedding, and audio models locally, with support for NVIDIA, AMD, Intel GPUs, CPUs, and Vulkan. It also provides a Python library and free notebooks for fast model fine-tuning.

## Use cases
- fine-tune llama or qwen models locally
- run deepseek and gemma models on my own gpu
- train a stable diffusion or flux image model
- fine-tune an llm with a gui without writing code
- run reinforcement learning training for llms
- self-host a chatgpt-like assistant
- fine-tune text-to-speech models

## When to choose
- you want a desktop UI to run and train models locally
- you need fast fine-tuning with limited GPU memory
- you want multi-GPU and multi-vendor GPU support
- you prefer ready-made training notebooks

## When to avoid
- you need a headless server-side inference API at scale
- you only want cloud-hosted model training
- you need non-Python tooling

## Facets
- artifact type: application
- maturity: active
- function: llm-inference, llm-training, machine-learning, stable-diffusion, tts, agent-framework, gui
- domain: large-language-models, machine-learning, deep-learning, artificial-intelligence, image-processing, speech-processing
- platform: windows, python
- tags: fine-tuning, desktop-app, local-llm, diffusion-models, qlora, self-hosted, multi-gpu, macos, linux, desktop, gpu

## Member repositories
- unslothai/unsloth (main) score 94
- unslothai/notebooks (examples) score 66

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:21.483761+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:14:48.053012+00:00, confidence not recorded.
  - readme: https://github.com/unslothai/unsloth (fetched 2026-08-28T04:12:21.483761+00:00, sha a3536339916e)
- Data as of 2026-08-30T08:39:29.467469+00:00.