unsloth: A desktop app for running and training models locally.
What problem it solves
Running and training models locally requires tools for model execution, fine-tuning, agent integration, and deployment across different model formats and hardware. The project brings these workflows into a desktop app, a web UI, and a code-based version.
Key features
Run and train models — Run and train language, diffusion, decision, embedding, audio, image, and video models through local workflows.
Source: README
Connect local models to agents — Use local models with coding agents and MCP, including tool calling and code execution.
Source: README
Fine-tune and export models — Train models with reinforcement learning, LoRA, QLoRA, full fine-tuning, pretraining, DPO, GRPO, and FP8, then export them in supported deployment formats.
Source: README
Search, retrieval, and browser tools — Combine private web search, deep research, RAG, browser use, and sandboxed command execution with local model conversations.
Source: README
Serve models locally or remotely — Expose models over LAN or secure remote access, or serve them through an OpenAI-compatible API.
Source: README
Quick start
Copied verbatim from the project's README.
curl -fsSL https://unsloth.ai/install.sh | shunsloth studiounsloth start claude --model unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_XLWho it may suit
This may suit developers and model users who want a graphical workflow for local model execution and training, as well as Python users who prefer the code-based version.
Topics to discuss
Which local model workflows should be prioritized in the desktop experience?
How should agent access and sandbox boundaries be designed for local models?
Which deployment and dataset-building workflows need clearer documentation?
Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.
Read the original README https://github.com/unslothai/unsloth#readme
ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.
Comments
…