Train AI models with Unsloth and Hugging Face Jobs for FREE

Fine-tuning large language models is now significantly faster and more affordable through the integration of Unsloth with Hugging Face Jobs. The Unsloth library dramatically speeds up training by about two times while cutting VRAM usage by approximately 60%, making it possible to train models for just a few dollars. This process is made incredibly accessible through a new skill for coding agents like Claude Code, allowing you to kick off a training job with a simple natural language prompt. The approach is particularly effective for smaller models, which are not only cheap to train but also optimized for deployment on everyday devices like CPUs and phones.