🛠️Unsloth Requirements

Here are Unsloth's requirements including system and GPU VRAM requirements.

Unsloth can be used in two ways: through Unsloth Studio, the web UI, or through Unsloth Core, the original code-based version. Each has different requirements.

Unsloth Studio Requirements

Inference

Unsloth Studio Inference, Data Recipes, and Export work on macOS, Windows, and Linux with or without a GPU, including CPU-only setups.

Training

Unsloth Studio Training currently works on NVIDIA GPUs, with AMD, MLX, Intel support coming very soon. You can still use the original Unsloth Core to train on AMD and Intel devices. Python 3.11–3.13 is required.

Requirement	Linux / WSL	Windows
Git	Usually preinstalled	Installed by setup script (winget)
CMake	Preinstalled or sudo apt install cmake	Installed by setup script (winget)
C++ compiler	build-essential	Visual Studio Build Tools 2022
CUDA Toolkit	Optional; nvcc auto-detected	Installed by setup script (matched to driver)

Unsloth Core Requirements

• Operating System: Works on Linux and Windows

• Supports NVIDIA GPUs since 2018+ including Blackwell RTX 50 and DGX Spark

• Minimum CUDA Capability 7.0 (V100, T4, Titan V, RTX 20 & 50, A100, H100, L40 etc) Check your GPU! GTX 1070, 1080 works, but is slow.

• The official Unsloth Docker image unsloth/unsloth is available on Docker Hub

  • Docker

• Unsloth works on AMD and Intel GPUs (follow our specific guides). Apple/Silicon/MLX is in the works

• Your device should have xformers, torch, BitsandBytes and triton support.

• If you have different versions of torch, transformers etc., pip install unsloth will automatically install all the latest versions of those libraries so you don't need to worry about version compatibility.

Python 3.13 is supported!

Fine-tuning VRAM requirements:

How much GPU memory do I need for LLM fine-tuning using Unsloth?

A common issue when you OOM or run out of memory is because you set your batch size too high. Set it to 1, 2, or 3 to use less VRAM.

For context length benchmarks, see here.

Check this table for VRAM requirements sorted by model parameters and fine-tuning method. QLoRA uses 4-bit, LoRA uses 16-bit. Keep in mind that sometimes more VRAM is required depending on the model so these numbers are the absolute minimum:

Model parameters	QLoRA (4-bit) VRAM	LoRA (16-bit) VRAM
3B	3.5 GB	8 GB
7B	5 GB	19 GB
8B	6 GB	22 GB
9B	6.5 GB	24 GB
11B	7.5 GB	29 GB
14B	8.5 GB	33 GB
27B	22GB	64GB
32B	26 GB	76 GB
40B	30GB	96GB
70B	41 GB	164 GB
81B	48GB	192GB
90B	53GB	212GB
405B	237 GB	950 GB