Why we are featuring it
Train models without turning setup into an infrastructure project.
Unsloth brings execution, fine-tuning, search, RAG, and agent connections together in a desktop app for local models.
What is inside
- Supports LLM, diffusion, embedding, and audio models.
- Fine-tuning with less VRAM and multiple training methods.
- Connects Claude Code, Codex, MCP, and OpenAI-compatible APIs.
- Supports Windows, Linux, WSL, macOS, GPUs, and CPUs.