Build AI workflows that
run on your machine.
NodeTool Studio is the desktop app for builders who want their models, data, and outputs to stay local. Use Ollama, MLX, and GGUF models with zero per-token cost — and bring your own keys for any cloud provider when you need them.
Free · AGPL-3.0 · macOS, Windows, Linux · No account required · All downloads

Privacy, speed, and zero per-token cost.
Everything Studio does, it can do without a network connection. Your prompts, your files, and your models — all on your hardware.
Works fully offline
Once your local models are downloaded, you can disconnect the internet and keep building. Perfect for travel, secure environments, or air-gapped machines.
Your data never leaves the device
Workflows, assets, prompts, and model outputs stay on your disk. No telemetry, no opt-out — there is no cloud round-trip unless you explicitly call a remote API.
Run open-weight models locally
Ollama, MLX (Apple Silicon), GGUF/GGML, HuggingFace Transformers — all wired into the same nodes. Pick any open model with zero API cost.
Use your GPU to the fullest
NVIDIA CUDA on Windows/Linux, Metal on Apple Silicon. Image, video, and audio nodes go straight to the metal.
Own your model library
Built-in Model Manager downloads, organizes, and shares model files across workflows. Curate the exact stack you want.
Custom Python and TypeScript nodes
Drop in your own code, install Python packages in the bundled environment, or sandbox untrusted code in Docker. Studio is your full local AI lab.
What Studio asks of you.
Local-first means you take on a few things the cloud would otherwise handle. If any of these feel heavy, NodeTool Cloud runs the same workflows in your browser.
Hardware matters
Local LLMs need RAM and ideally a GPU. We recommend 16GB+ RAM and 4GB+ VRAM for serious local-model work.
You manage updates
When a new release ships, you install it. We sign and notarize builds for macOS/Windows so updates stay easy.
Disk space
Local models are large — plan ~20GB for a small starter set, much more for video/image weights.
100% open source. Always.
Studio is released under AGPL-3.0. Every node, every provider, every line of the runtime is on GitHub. Audit it, fork it, self-host it — there is no separate “pro” codebase.
View source on GitHub