Unsloth - Train and Run Models LocallyunslothModelsBlogUnsloth Studio✨Docs
New
✨Introducing Unsloth StudioEasily run & train models locally.
Join our DiscordStart for free
Latest News
Gemma 4 12B and QAT is here!Jun 5, 2026
Unsloth joins PyTorch ecosystemMay 11, 2026
Unsloth API endpointMay 5, 2026
Qwen3.6 is out now!Apr 22, 2026
View more news<br>Run models locally<br>Unsloth Studio runs 100% offline on your Mac and Windows device. Run GGUF and Safetensors models with tool-calling, web search, and OpenAI compatible API.
Compare models side by side and upload images, docs, audio, code files and more.<br>Learn more
No-code training<br>Auto-create datasets from PDF, CSV, JSON docs and start training with real-time observability.
Unsloth's custom kernels supports optimized training for LoRA, FP8, FFT, PT and 500+ models including text, vision, audio and embeddings.<br>QuickstartLearn more
Model Arena<br>Chat with and compare 2 different models, such as a base model and a fine-tuned one, to see how their outputs differ.
Just load your first GGUF/model, then the second, and voilà!<br>Learn more
Data Recipes<br>Data Recipes transforms your docs into useable datasets via graph-node workflow. Upload unstructured or structured files like PDFs, CSV and JSON. Unsloth Data Recipes auto turns documents into your desired formats.<br>QuickstartLearn more
Export models<br>Export any model, including your fine-tuned models, to safetensors, or GGUF for use with llama.cpp, vLLM, Ollama, and more.<br>Learn more
Don’t believe us?<br>Why not try our fully free open source version? Finetune 2X faster on a single NVIDIA GPU for free on Google Colab or Kaggle Notebooks.<br>Get access now
Sign up to our newsletter<br>We'll share monthly updates!Submit
Subscribe now
Train your own custom model in 24 hrs, not 30 days.
30x faster than FA2 + 30% accuracy
90% less memory usage than FA2
audio, embedding, vision support
The details<br>We're making AI more accessible to everyone<br>Find out moreUnsloth makes everything greener<br>As hardware costs rise and performance gains plateau, we use our math and coding skills to make models train and run smarter + faster.
Want lightning fast inference? We’re working on it!<br>Contact us
Don't forget to join our newsletter!<br>Submit<br>By registering you agree to unsloth's Terms of Service and Privacy Policy,<br>Subscribe now
MultiGPU DocsEven better multiGPU in the works!
Don't forget to join our newsletter!<br>Subscribe
Pricing<br>FreeFreeware of our standard version of unsloth<br>Get started<br>Open-source
Supports Mistral, Gemma
Supports LLama 1, 2, 3
MultiGPU - coming soon
Supports 4 bit, 16 bit LoRA
unsloth Pro2.5x faster training + 20% less VRAM<br>Contact us<br>2.5x number of GPUs faster than FA2
20% less memory than OSS
Enhanced MultiGPU support
Up to 8 GPUS support
For any usecase
unsloth EnterpriseUnlock 30x faster training + multi-node support + 30% accuracy<br>Contact us<br>32x number of GPUs faster than FA2
up to +30% accuracy
5x faster inference
Supports full training
All Pro plan features
Multi-node support
Customer support
Ready to use unsloth?<br>Get started for free
Company<br>About📰 NewsletterPrivacy PolicyTerms of Service<br>Product<br>Introduction🐋 DockerDownloadDocumentation🦥 Models
Community
Twitter (X)
Hugging Face
Discord
unsloth<br>[email protected]
© 2026 unsloth. All rights reserved.
Join Our Discord