About
Orchard is a native Swift application for macOS that lets you manage containers, machines, and local AI models built on Apple's container tooling. It provides a desktop experience that complements the container command-line interface, and is presented as a replacement for Docker Desktop.
For local AI, Orchard connects MLX models to Apple containers. Inference runs on your Mac's GPU, and containers can reach model servers without hand-configured networking. The app can discover model servers already running on your Mac (such as Ollama, LM Studio, and MLX servers), or start and stop mlx_lm.server instances itself. It also includes a sandboxes view for isolated workloads, showing network mode, mounts, resource limits, and ownership, with terminal access, chat when a model is attached, and a kill-switch, plus an in-app chat tester for verifying model servers.
Orchard also manages Apple container machines, which are persistent, stateful Linux VMs, letting you create, configure, start, stop, and monitor them with live resource usage and logs. It further manages local Kubernetes clusters built on Apple container's cluster plugin, grouping node containers and providing a one-click kubectl setup. Orchard is installable via Homebrew or a release download, and is open source under the MIT license.
What's new
All release notesHomebrew cask orchard 2.6.0 (sha256 64c4e04d0a7dc1daffaa5cf76e8e00f6dd70772e75a80167936e53a9f88afb91).