unslothai/unsloth at 65K Stars: Training UI for Open Models Now Supports Gemma 4, Qwen 3.6, DeepSeek, and GPT-OSS Locally

GitHub·medium signal

Unsloth Studio, the web UI for training and running open models, is trending at 65K stars with expanded model support including Gemma 4, Qwen 3.6, DeepSeek, and GPT-OSS. The platform provides fine-tuning capabilities with 2-5x faster training speeds and 80% less memory usage through custom CUDA kernels, making local model training accessible to developers without deep ML expertise. The addition of GPT-OSS support signals the broadening open-weights ecosystem.