unsloth
Local UI to run and train LLMs and diffusion models, including Qwen3.8, Kimi K3, MiniMax-H3, Gemma 4, DeepSeek-V4, FLUX and more.
Unsloth is a desktop application and AI sub-agent platform for running and training large language models, diffusion models, and other AI architectures locally. It enables developers to fine-tune LLMs up to 2× faster with 70% less VRAM, supports models like Qwen3.8, Kimi K3, Gemma 4, DeepSeek-V4, and FLUX, and integrates with coding agents such as Claude Code and Codex. Available for Windows, macOS, and Linux, it offers both a graphical interface (Unsloth Desktop/Studio) and code-based workflows.
Key Features
Use Cases
- 01Fine-tuning open-source LLMs on custom datasets for domain-specific tasks without cloud dependencies
- 02Running AI coding assistants like Claude Code or Codex with locally hosted models for privacy-sensitive development
- 03Training diffusion models for image and video generation on consumer GPUs with optimized memory usage
- 04Building local RAG systems with private web search and document processing capabilities
- 05Deploying lightweight LLMs on edge devices or private infrastructure via GGUF quantization
- 06Experimenting with reinforcement learning and vision-language models using free Colab notebooks
Related Agents
View morecodegraph
Pre-indexed code knowledge graph, auto syncs on code changes, for Claude Code, Codex, Gemini, Cursor, OpenCode, AntiGravity, Kiro, and Hermes Agent — fewer tokens, fewer tool calls, 100% local
claude-code-best-practice
from vibe coding to agentic engineering - practice makes claude perfect
system_prompts_leaks
Extracted system prompts from Anthropic - Claude Fable 5, Opus 5, Claude Design, Claude Code. OpenAI - ChatGPT GPT-5.6-Sol, Codex. Google - Gemini 3.5 Flash, 3.1 Pro, Antigravity. xAI - Grok, Cursor, Copilot, VS Code, Perplexity, and more. Updated regularly.
impeccable
The design language that makes your AI harness better at design.
unsloth — FAQ
What is Unsloth?+
Unsloth is a local AI platform for running and training large language models, diffusion models, and other AI architectures on your own hardware. It provides desktop and web interfaces plus integration with popular coding agents like Claude Code and Codex.
How do I install Unsloth?+
Download the native desktop app for Windows, macOS, or Linux from the GitHub releases page, or install Unsloth Studio via one-line shell command (curl -fsSL https://unsloth.ai/install.sh | sh on Mac/Linux, or irm https://unsloth.ai/install.ps1 | iex on Windows). For code-based workflows, install Unsloth Core via pip using uv or standard Python package managers.
Which AI clients and agents work with Unsloth?+
Unsloth integrates with Claude Code, OpenAI Codex, Hermes Agent, OpenClaw, and OpenCode through the 'unsloth start' command. It exposes OpenAI-compatible and Anthropic-compatible APIs, allowing any client that supports these standards to connect to locally hosted models.
Do I need API keys to use Unsloth?+
No external API keys are required to run or train models locally with Unsloth. However, if you enable cloud provider connections or use features like web search integrations, those services may require their own credentials.
Is Unsloth free to use?+
Yes, Unsloth is open-source and free to use. The software, desktop app, and training notebooks are available at no cost, though you need compatible hardware (NVIDIA, AMD, Intel, or Apple Silicon GPUs) to run models efficiently.
What hardware does Unsloth support?+
Unsloth supports NVIDIA GPUs (RTX 30/40/50, Blackwell, DGX), AMD GPUs (RDNA 2+, Vega via ROCm), Intel GPUs (via Vulkan), Apple Silicon Macs (via Metal/MLX), and CPU-only inference. Multi-GPU configurations and mixed precision (FP8, GGUF) are also supported.
How do I install unsloth?+
Open the source repository on GitHub and follow its README. unsloth is a agent — MCP Agents Market links you directly to the official repo.
Is unsloth free?+
unsloth is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.