Unsloth Studio Installation

Learn how to install Unsloth Studio on your local device.

Unsloth Studio works on Windows, Linux, WSL and MacOS. You should use the same installation process on every device, although the system requirements may differ by device.

WindowsMacOSLinux & WSLDockerDeveloper Install

• Mac: Like CPU - Chat + Data Recipes works for now. MLX training coming very soon.

• CPU: Unsloth still works without a GPU, but for Chat + Data Recipes.

• Training: Works on NVIDIA: RTX 30, 40, 50, Blackwell, DGX Spark/Station etc. + Intel GPUs

• Coming soon: Support for Apple MLX and AMD.

Install Instructions

Remember install instructions are the same across every device:

1

Install Unsloth

MacOS, Linux, WSL:

curl -fsSL https://unsloth.ai/install.sh | sh

Windows PowerShell:

irm https://unsloth.ai/install.ps1 | iex

First install should now be 6x faster and with 50% reduced size due to precompiled llama.cpp binaries.

WSL users: you will be prompted for your sudo password to install build dependencies (cmake, git, libcurl4-openssl-dev).

2

Launch Unsloth Studio

unsloth studio -H 0.0.0.0 -p 8888

Then open http://localhost:8888 in your browser.

3

Onboarding

On first launch you will need to create a password to secure your account and sign in again later. You’ll then see a brief onboarding wizard to choose a model, dataset, and basic settings. You can skip it at any time.

4

Start training and running

Start fine-tuning and building datasets immediately after launching. See our step-by-step guide to get started with Unsloth Studio:

Get Started

Update Unsloth Studio:

Use the same install commands to update.

MacOS, Linux, WSL:

curl -fsSL https://unsloth.ai/install.sh | sh

Windows PowerShell:

irm https://unsloth.ai/install.ps1 | iex

Or use (currently does not work on Windows):

unsloth studio update 

System Requirements

Windows

Unsloth Studio works directly on Windows without WSL. To train models, make sure your system satisfies these requirements:

Requirements

• Windows 10 or Windows 11 (64-bit)

• NVIDIA GPU with drivers installed

• App Installer (includes winget): here

• Git: winget install --id Git.Git -e --source winget

• Python: version 3.11 up to, but not including, 3.14

• Work inside a Python environment such as uv, venv, or conda/mamba

MacOS

Unsloth Studio works on Mac devices for Chat for GGUF models and Data Recipes (Export coming very soon). MLX training coming soon!

• macOS 12 Monterey or newer (Intel or Apple Silicon)

• Install Homebrew: here

• Git: brew install git

• cmake: brew install cmake

• openssl: brew install openssl

• Python: version 3.11 up to, but not including, 3.14

• Work inside a Python environment such as uv, venv, or conda/mamba

Linux & WSL

• Ubuntu 20.04+ or similar distro (64-bit)

• NVIDIA GPU with drivers installed

• CUDA toolkit (12.4+ recommended, 12.8+ for blackwell)

• Git: sudo apt install git

• Python: version 3.11 up to, but not including, 3.14

• Work inside a Python environment such as uv, venv, or conda/mamba

Docker

Our Docker image now works for Studio! We're working on Mac compatibility.

• Pull our latest Unsloth container image: docker pull unsloth/unsloth

• Run the container via:

docker run -d -e JUPYTER_PASSWORD="mypassword" \
  -p 8888:8888 -p 8000:8000 -p 2222:22 \
  -v $(pwd)/work:/workspace/work \
  --gpus all \
  unsloth/unsloth

For more information, see here.

• Access your studio instance at http://localhost:8000 or external ip address http://external_ip_address:8000/

CPU only

Unsloth Studio supports CPU devices for Chat for GGUF models and Data Recipes (Export coming very soon)

• Same as the ones mentioned above for Linux (except for NVIDIA GPU drivers) and MacOS.

Developer Installation (Advanced)

macOS, Linux, WSL developer installs:

curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv unsloth_studio --python 3.13
source unsloth_studio/bin/activate
uv pip install unsloth --torch-backend=auto
unsloth studio setup
unsloth studio -H 0.0.0.0 -p 8888

Windows PowerShell developer installs:

winget install -e --id Python.Python.3.13
winget install --id=astral-sh.uv  -e
uv venv unsloth_studio --python 3.13
.\unsloth_studio\Scripts\activate
uv pip install unsloth --torch-backend=auto
unsloth studio setup
unsloth studio -H 0.0.0.0 -p 8888

Nightly - MacOS, Linux, WSL:

curl -LsSf https://astral.sh/uv/install.sh | sh
git clone --filter=blob:none https://github.com/unslothai/unsloth.git unsloth_studio
cd unsloth_studio
uv venv --python 3.13
source .venv/bin/activate
uv pip install -e . --torch-backend=auto
unsloth studio setup
unsloth studio -H 0.0.0.0 -p 8888

Then to launch every time:

cd unsloth_studio
source .venv/bin/activate
unsloth studio -H 0.0.0.0 -p 8888

Nightly - Windows:

Run in Windows Powershell:

winget install -e --id Python.Python.3.13
winget install --id=astral-sh.uv  -e
git clone --filter=blob:none https://github.com/unslothai/unsloth.git unsloth_studio
cd unsloth_studio
uv venv --python 3.13
.\.venv\Scripts\activate
uv pip install -e . --torch-backend=auto
unsloth studio setup
unsloth studio -H 0.0.0.0 -p 8888

Then to launch every time:

cd unsloth_studio
.\.venv\Scripts\activate
unsloth studio -H 0.0.0.0 -p 8888

Uninstall

You can uninstall Unsloth Studio by deleting its install folder usually located under $HOME/.unsloth/studio on Mac/Linux/WSL and %USERPROFILE%\.unsloth\studio on Windows. Or run:

• MacOS, WSL, Linux: rm -rf ~/.unsloth/studio

• Windows (PowerShell): Remove-Item -Recurse -Force "$HOME\.unsloth\studio"

• Optional: remove $HOME\.unsloth on Windows or ~/.unsloth on MacOS/Linux/WSL if you want to delete all Unsloth files

Note: Using the rm -rf commands will delete everything, including your history, cache, chats etc.

Deleting model files

You can delete old model files either from the bin icon in model search or by removing the relevant cached model folder from the default Hugging Face cache directory. By default, Hugging Face uses ~/.cache/huggingface/hub/ on macOS/Linux/WSL and C:\Users\<username>\.cache\huggingface\hub\ on Windows.

• MacOS, Linux, WSL: ~/.cache/huggingface/hub/

• Windows: %USERPROFILE%\.cache\huggingface\hub\

If HF_HUB_CACHE or HF_HOME is set, use that location instead. On Linux and WSL, XDG_CACHE_HOME can also change the default cache root.

Google Colab notebook

We’ve created a free Google Colab notebook so you can explore all of Unsloth’s features on Colab’s T4 GPUs. You can train and run most models up to 22B parameters, and switch to a larger GPU for bigger models. Just Click 'Run all' and the UI should pop up after installation.

Google Colabcolab.research.google.com

Once installation is complete, scroll to Start Unsloth Studio and click Open Unsloth Studio in the white box shown on the left:

Scroll further down, to see the actual UI.

We now precompile llama.cpp binaries for much faster install speeds.

Sometimes the Studio link may return an error. This happens because you might be using an adblocker or Mozilla or Google Colab expects you to stay on the Colab page; if it detects inactivity, it may shut down the GPU session. Nevertheless, you can scroll down a bit

Troubleshooting

Problem	Fix
Python version error	sudo apt install python3.12 python3.12-venv version 3.11 up to, but not including, 3.14
nvidia-smi not found	Install NVIDIA drivers from https://www.nvidia.com/Download/index.aspx
nvcc not found (CUDA)	sudo apt install nvidia-cuda-toolkit or add /usr/local/cuda/bin to PATH
llama-server build failed	Non-fatal, Studio still works, GGUF inference won't be available. Install cmake and re-run setup to fix.
cmake not found	sudo apt install cmake
git not found	sudo apt install git
Build failed	Delete ~/.unsloth/llama.cpp and re-run setup