Unsloth Studio: Web UI for Local Training and Running of Gemma 4, Qwen3.5, and DeepSeek Hits 59.5K Stars

GitHub·medium signal

Unsloth (59,478 stars) now offers Unsloth Studio, a web UI for training and running open models locally including the newly released Gemma 4 (all 4 sizes) and Qwen3.5 (including 397B-A17B). Trains 500+ models up to 2x faster with 70% less VRAM and no accuracy loss. Studio supports Windows, Linux, WSL, and macOS with NVIDIA RTX 30/40/50 and Blackwell GPU training. Includes tool calling, code execution in sandboxes, and LoRA/QLoRA fine-tuning.