Anomaly Brand, the company behind OpenCode, launched a public desktop beta of the open-source AI coding agent on April 2, 2026, making the tool available as a native application for macOS, Windows, and Linux. The release is available for download at opencode.ai. Author details for the engineering team were not available at time of publication.
- OpenCode desktop beta is now live for macOS, Windows, and Linux, joining the existing terminal interface and IDE extension.
- The platform connects to more than 75 LLM providers through Models.dev, including Claude, GPT, Gemini, and locally hosted models.
- Existing GitHub Copilot and ChatGPT Plus/Pro subscribers can authenticate directly, with no separate OpenCode subscription required.
- OpenCode states it does not retain code or context data on its servers, positioning it for use in privacy-sensitive development environments.
What Happened
Anomaly Brand released the OpenCode desktop beta for macOS, Windows, and Linux, expanding the open-source AI coding agent beyond its original terminal and IDE extension interfaces into a standalone native application available at no cost. Prior to this release, OpenCode’s interfaces were limited to the terminal application and an IDE extension; the new desktop app provides a graphical environment without requiring a terminal or IDE to be running. According to the company’s website, OpenCode is already “used and trusted by over 5 million developers every month,” with the GitHub repository surpassing 120,000 stars at time of publication.
Why It Matters
OpenCode’s desktop beta places it in direct competition with proprietary AI coding tools in a market where most prominent options are either vendor-locked or bundled with specific inference layers. GitHub Copilot operates within Microsoft’s product ecosystem; tools such as Cursor bundle their own model access by default. OpenCode treats model selection as a developer-configurable variable rather than a product-level decision, supporting more than 75 providers simultaneously.
The project’s community metrics distinguish it from single-organization efforts: 800 contributors and more than 10,000 commits indicate distributed, ongoing development rather than a single vendor’s release cycle. An open-source codebase allows security-conscious organizations to audit data handling behavior directly — a meaningful factor for enterprises with formal approval requirements before deploying developer tooling.
Technical Details
OpenCode connects to more than 75 LLM providers through Models.dev, which aggregates cloud services including Claude, GPT, and Gemini alongside regional providers and locally hosted models, giving developers the option to route inference entirely within their own infrastructure. The platform also supports direct authentication with GitHub Copilot accounts and with ChatGPT Plus or Pro accounts, allowing developers holding existing subscriptions to either service to use OpenCode without purchasing an additional plan.
A core technical feature is automatic Language Server Protocol (LSP) loading, which provisions the appropriate language-specific tooling for the active LLM without requiring manual configuration — making type checking, reference tracking, and autocomplete context available to the model at generation time. Multi-session support allows several OpenCode agents to run in parallel on the same project, enabling concurrent workstreams such as writing tests in one session while refactoring a module in another.
A session-sharing feature generates a permanent link for any active or completed agent session, providing a reference URL shareable with teammates for debugging or collaborative review. Installation is available through five package managers: curl via a bash script hosted at opencode.ai/install, npm, bun, brew, and paru.
Who’s Affected
Developers working across multiple AI providers, or those who need to swap models between projects with different performance or cost requirements, benefit most directly from OpenCode’s provider-neutral architecture. Teams subject to organizational data policies will find OpenCode’s stated privacy posture relevant: the company’s website asserts that the tool “does not store any of your code or context data, so that it can operate in privacy sensitive environments.” This is a stated product property, not an independently audited certification — organizations requiring verified compliance would need to conduct their own assessments.
Developers already paying for GitHub Copilot or ChatGPT Plus/Pro subscriptions are a specific target group: OpenCode enables them to reuse those existing credentials rather than paying a third subscription fee. The desktop beta also addresses developers who prefer a native application environment over terminal-based tooling, a preference that prior versions of OpenCode could not accommodate.
What’s Next
Anomaly Brand operates a companion service called Zen, described on the OpenCode website as providing “access to a handpicked set of AI models that OpenCode has tested and benchmarked specifically for coding agents.” Zen is positioned for developers who prefer a pre-validated model shortlist over self-selecting from the full 75-plus provider catalogue; the service targets inconsistencies in output quality across the broader provider landscape, though no specific benchmark results or testing methodology are published on the website at time of publication.
The desktop application remains in beta status. Anomaly Brand has opened a waitlist for additional undisclosed product launches. Feature development and roadmap activity are visible through the project’s public GitHub repository, which has logged more than 800 contributors and 10,000 commits and remains the primary channel for tracking OpenCode’s progress.
