Run Flow on your own machine
Flow is in active pre-release. Pick an edition below. Each one runs every model locally, so nothing you build leaves your machine. Packaged installers will appear here as they ship. Until then, each card links to its install guide.
Four ways to install
One Rust orchestration core powers all four surfaces. Install the one that fits where you work. Your flows and catalogs move between them unchanged.
Flow Studio
Pre-releaseThe desktop edition gives you a multi-tab canvas, Model Hub, Template Hub, and agentic runs, zero-egress by default. Packaged installers will appear here as they ship. For now, the install guide covers building and running it today.
Install guideFlow Code
ShippedA native VS Code sidebar where your selected local model generates a Flow DSL and the orchestrator runs it in-editor while the model monitors the run. Each tab picks its own model with a run-config popover for context size and reasoning, runs are editable and export to DSL or output, settings are shared with every other edition, and a context-usage indicator shows how much of the window you are using. The engine owns every file and shell action. A JetBrains plugin that re-hosts the same edition is planned.
Set up in VS CodeFlow Server
AvailableThe same platform on your own private cloud instance, served to a browser client. Inference runs inside the instance and is never sent to a third-party model service.
Deploy guideFlow CLI
PreviewFlow CLI is a headless runner and full-screen TUI. It executes Flow DSL directly on the shared core, and it is built for terminals and CI pipelines.
CLI installFlow is local-first by architecture. Models download once and run on your hardware, and execution authority stays with the engine. No operational data is sent to a third-party service.
Not sure which edition?
The same flows run everywhere. Compare the surfaces and execution loci, then install the one that fits.