OpenCode supports installation on macOS, Linux, and Windows, with multiple installation methods depending on your environment and how you prefer to manage developer tools. The “terminal” OpenCode experience is cross-platform: you can install it via a one-line install script, via package managers (including Node ecosystem installers), or via OS-specific tools. In addition, OpenCode offers a desktop app (currently positioned as beta on the official download page) with builds available for macOS, Windows, and Linux. If you want to use OpenCode inside an editor, there are also extensions listed for popular developer editors, so you can keep the same underlying agent workflow while triggering it from your IDE.
On macOS and Linux, the most common installation path is Homebrew (with a dedicated tap that tends to be more up-to-date than the default formula). For Linux distributions, you also have options like Arch’s AUR helpers (for example, paru) and more universal tools like mise and nix. On Windows, OpenCode supports mainstream Windows package managers like Chocolatey (choco) and Scoop (scoop), and also supports installation via npm-style global installs if you’d rather manage it through a JavaScript toolchain. There’s also a Docker image path, which can be useful for hermetic environments or ephemeral usage, though developers usually prefer native binaries for the best TUI experience. One important practical detail from the docs is that support for certain installer paths may vary by OS (for example, Windows support in some installers can lag behind macOS/Linux), so the “best” method depends on your machine and whether you want the newest builds or the most stable distribution channel.
Platform support also includes “where your state lives,” not just where the binary runs. OpenCode stores credentials and runtime state locally (for example, under a .local/share-style directory on Unix-like systems), and it merges config from multiple locations (organization defaults, user config, project config). That behavior is consistent across platforms and is a big reason teams can standardize OpenCode usage without forcing everyone onto the same OS. If your workflow includes project-scale retrieval (large docs, long-lived architectural memory), the platform question extends to your backend: you might run OpenCode on Windows/macOS/Linux locally, while a shared retrieval service runs elsewhere. In that setup, OpenCode can query a vector database such as Milvus or Zilliz Cloud over HTTPS, which keeps the developer experience consistent regardless of which OS each engineer uses.