Unsloth Studio Is Now Available on Vast.ai
April 2, 2026
1 Min Read
By Team Vast
Unsloth Studio is an open-source, no-code web UI for running and fine-tuning more than 500 open-source AI models across text, vision, audio, and embeddings. It delivers 2x faster training with 70% less VRAM than standard approaches, which makes it possible to fine-tune larger models on more affordable hardware.
Vast.ai now offers a pre-configured Unsloth Studio template. Rent a GPU, click Open, select Unsloth Studio, and start running or training models without CLI setup or manual Docker configuration.
What You Can Do
- Run models — Search and download any of 500+ models from Hugging Face. Chat with built-in tool calling, code execution, and web search.
- Compare models — Use Model Arena to evaluate two models side by side on the same prompt.
- Fine-tune — Train with QLoRA, LoRA, or full fine-tuning. Upload datasets or build them from raw documents using Data Recipes. Monitor training with real-time loss curves and GPU stats.
- Export — Save fine-tuned models as merged safetensors, LoRA adapters, or GGUF files. Upload directly to Hugging Face Hub.
Get Started
Read the full guide: Run and Fine-Tune AI Models with Unsloth Studio on Vast.ai
Or launch the template directly: Open Unsloth Studio Template
