Unsloth
Unsloth is a desktop application and toolkit for running and fine-tuning large language and diffusion models locally, supporting formats such as GGUF and MLX across NVIDIA, AMD and Intel GPUs as well as CPUs.
It runs and trains language, image and audio models, connects local models to coding agents such as Claude Code and Codex with tool calling over MCP, and adds private web search, deep research and a rolling-context RAG mode. Fine-tuning is reported to run twice as fast with about 70 percent less VRAM than standard methods, with support for LoRA, QLoRA, full fine-tuning, pretraining and reinforcement-learning methods including GRPO and DPO, plus export to formats such as GGUF and NVFP4.
A one-command "Unsloth Start" connects Claude Code, Codex and other agents to a chosen local model, and a served OpenAI-compatible API lets a locally hosted model be reached remotely over HTTPS or from another application.
Available as a desktop app, a web UI called Unsloth Studio, or a code-based core library.