Unsloth Desktop is an open-source app to run and train AI models locally. Run LLMs, image/video diffusion, and audio. Connect agents like Claude Code or Codex to your local GPU with one command, and fine-tune models with no-code workflows.
Loading comments…
Achievement
Project Info
Product Keywords
Unsloth Desktop is an open-source application that brings both AI inference and model training to your local machine. It runs fully offline on Mac and Windows, letting you load and chat with LLMs, generate images and video through diffusion models, and work with audio — all without sending data to the cloud. Beyond running models, it includes no-code fine-tuning workflows, dataset creation tools, and a model comparison arena, making it a complete local AI workbench rather than just another chat client.
Unsloth Desktop supports both GGUF and Safetensors formats, with tool-calling, web search, and an OpenAI-compatible API built in. You can upload images, docs, audio, and code files directly into the chat context, and everything runs 100% offline on your device.
Create datasets automatically from PDF, CSV, or JSON files, then start fine-tuning with real-time observability. The underlying custom kernels support optimized training for LoRA, FP8, FFT, and PT across 500+ models — including text, vision, audio, and embeddings.
Chat with two different models side by side — say, a base model and its fine-tuned version — to compare outputs directly. Load your first model, load the second, and see exactly how their responses differ in real time.
Transform unstructured or structured documents into usable datasets through a graph-node workflow. Upload PDFs, CSVs, or JSON files, and Unsloth Desktop automatically converts them into the formats you need for training.
Unsloth Desktop turns a single local GPU into a full AI studio — run, compare, and fine-tune models without writing a line of code.
The combination of inference, training, and dataset tooling in one offline app is rare. Most local AI tools focus on one side of the equation — either running models or training them — but Unsloth Desktop bridges both. The performance claims are also notable: up to 30x faster training than FlashAttention-2 with 90% less memory usage, plus support for audio, embedding, and vision models. And because it's open-source, you can start with the free version and upgrade to Pro or Enterprise only if you need multi-GPU or multi-node support.
You want a single desktop app that handles both running and training AI models locally, especially if you value privacy, prefer visual workflows over scripting, or need to connect coding agents to a local GPU without cloud dependencies. It's also a strong fit if you're experimenting with fine-tuning and want to compare base models against your custom versions side by side.
Other tools you might consider
The world can't build compute fast enough to keep up with AI demand. So we took a different path. ZeroGPU is AI infrastructure powered by small language models running on a hybrid edge network reusing compute that already exists. Not every task needs a frontier model. Our purpose-built, edge-optimized models run 10x faster, 50% cheaper and offload 70–80% of production tasks to small models with frontier-level accuracy.
Run state-of-the-art open-source models (GLM 5.1, Kimi K2.7 Code, MiniMax M2.7, and more) in Claude Code at up to 4× the speed (up to 200 tok/s) for a flat $29/month. Set up in minutes, no code changes.
nullsink provides anonymous access to Claude, GPT, and confidential open models without requiring an account, email, card, or subscription. Fund a bearer API key from $2 using Monero or Bitcoin, then use it with native Anthropic or OpenAI-compatible clients. Credit does not expire. nullsink keeps no request-content or client-IP logs and publishes its AGPL source code, architecture, and trust-model limitations.
GitHits gives coding agents access to the open-source code your app depends on. Get real implementation examples, dependency source navigation, package inspection and documentation. Agents can grep and read your codebase. They can't grep and read the open-source code your app depends on. That's where they start guessing, retrying, and looping. GitHits builds a version-aware index on demand. Agents can search, navigate, and inspect the code behind their dependencies. CLI: npx githits@latest init
Maker
calm_kit
Compare with
Alternatives
Loading comments…