Unsloth Adds Gemma 4, Qwen3.6, and gpt-oss Support — Web UI for Training Open Models Locally at 63.2K Stars

GitHub·medium signal

Unsloth, the popular open-source tool for fine-tuning and running LLMs locally, has updated its Web UI to support the latest generation of open models including Google's Gemma 4, Alibaba's Qwen3.6, and the new gpt-oss models alongside existing DeepSeek support. At 63,162 stars with active pushes today, Unsloth's 2x faster training with 70% less memory usage makes it the go-to tool for developers wanting to fine-tune frontier open-weight models without cloud GPU costs.