Unsloth Studio: Open-Source Web UI to Train and Run LLMs Locally

Melvin Vivas · X video post · 2026-03-17 · 0:45 · 192 views · Open on X

Topics: Fine-tuning & Model Customization, AI Dev Tools & Productivity · Level: beginner

Summary

Melvin Vivas shares the launch of Unsloth Studio, an open-source web UI from the Unsloth team for training (fine-tuning) and running LLMs on your own machine. The quoted announcement says it works on Mac, Windows and Linux, makes training faster while using less VRAM, supports several model types, and can build training datasets from your documents. The short video shows that it can run audio models.

Key points

• Unsloth Studio is a new open-source web UI for both training and running LLMs locally.
• It runs on Mac, Windows and Linux.
• The announcement says it can train 500+ models 2x faster while using 70% less VRAM.
• It supports GGUF models as well as vision, audio and embedding models.
• It can automatically create training datasets from PDF, CSV and DOCX files.
• The video shows Unsloth Studio running an audio model.
• A GUI lowers the barrier to local fine-tuning compared with writing training scripts by hand.

Resources mentioned

• Unsloth · tool · unsloth.ai · free
  Open-source library for fast, memory-efficient fine-tuning of open LLMs (LoRA/QLoRA) on a single GPU or in Colab.
  Also in: Run Laya Decision models locally with Unsloth on 4GB RAM (Melvin Vivas on X · notes), Unsloth passes 500M model downloads on Hugging Face (Melvin Vivas on X · notes), Run Qwen-Image-2.1 locally on 12GB VRAM with Unsloth GGUFs (Melvin Vivas on X · notes), Base vs fine-tuned Gemma 4 E2B as a model router (Melvin Vivas on X · notes) and 29 more