Coder

Coder is a self-hosted platform for development workspaces and AI coding agents, aimed at teams that want to run both on infrastructure they control. Its main appeal is the combination of customizable workspaces and delegated agent tasks; the trade-off is taking responsibility for a self-hosted environment.

Overview

Workspaces are defined with Terraform and connect through a secure WireGuard tunnel. Developers can reach them from VS Code, JetBrains, Cursor, a web terminal, remote desktop, or SSH, so teams can keep varied editor and access preferences without splitting across workspace platforms. Idle workspaces can stop automatically and restart when needed, a practical way to avoid leaving unused environments running.

Coder Agents can take on development or research work through chat or an API, provisioning a workspace to execute each task. The agent loop runs in the control plane, keeping LLM credentials out of workspaces. Support for Anthropic, OpenAI, Google, Bedrock, and self-hosted model endpoints gives teams provider choice, while the self-hosted approach makes this best suited to organizations prepared to manage their own infrastructure.

Key features

• Custom workspace environments: Terraform-defined configuration, persistent storage, prebuilds, GPU workspaces, and multiple VCS hosts give teams room to shape environments around their development needs. That flexibility is valuable for established teams, but it also means Coder is not simply a ready-made hosted editor.
• Broad access options: VS Code, JetBrains, Cursor, browser terminal, remote desktop, and SSH cover different workflows. The CLI is available on Linux, macOS, and Windows.
• Workspace lifecycle controls: Automatic stopping and restarting of idle workspaces can reduce wasted runtime while keeping environments available when needed.
• Delegated agent work: Agents receive development or research tasks through chat or API and work in provisioned environments. Keeping the agent loop in the control plane separates LLM credentials from the workspace.
• Integrations and security posture: Coder documents connections to GitHub, GitLab, Bitbucket, and JFrog Artifactory. Its site displays a SOC 2 Type II Certified badge.

Pricing

Community — free includes unlimited self-hosted workspaces, up to five active Coder Agents, and one platform integration. Support is through GitHub and Discord. This is a strong starting point for teams that can operate their own deployment and stay within the agent and integration limits; it does not include the governance, availability, and support commitments of Premium.

Premium — custom pricing adds advanced governance, high availability and workspace proxies, OIDC synchronization, audit logging, SLA-backed support, and ticket-based global support. It retains the limit of up to five active Coder Agents, so it suits organizations prioritizing operational controls and support rather than a larger agent allowance.

AI Premium — custom pricing includes self-hosted AI workflows, the Coder Agents API, network-isolated agent workspaces, model connections for Anthropic, OpenAI, Azure, and AWS Bedrock, and unlimited agents. It is the relevant upgrade for teams whose agent workload exceeds five active agents or that need the listed isolated workflow capabilities. Coder offers a free trial, but no trial length or renewal terms are stated.

Platforms

Coder supports API, Linux, macOS, self-hosted, web, and Windows. Self-hosting is central to the product rather than an optional deployment detail; teams seeking a fully managed workspace service may prefer a different approach.

Who it's for

Coder is best for teams that want control over workspace infrastructure, need custom environments, or want to delegate coding and research inside provisioned workspaces. Its editor breadth, persistent storage, prebuilds, GPU workspaces, and multiple VCS hosts support varied development setups. It is less compelling for individuals who only need a simple browser-based coding space or teams unwilling to own self-hosted operations.

Pros and cons

• Pro: Unlimited Community workspaces at no cost make it possible to build a substantial self-hosted setup without a workspace-count cap.
• Pro: The control-plane agent loop keeps LLM credentials out of workspaces, addressing a meaningful security concern for delegated work.
• Pro: Access through several editors, terminal, desktop, and SSH avoids forcing developers into one interface.
• Con: Community is capped at five active agents and one platform integration, which can constrain broader adoption.
• Con: Premium still caps active agents at five; teams needing more must consider AI Premium.
• Con: Self-hosting offers control but requires the team to take responsibility for its infrastructure.

Alternatives

Choose GitHub Codespaces if you prefer usage-billed cloud compute, with listed 2-core and 4-core compute plans, over operating self-hosted workspaces. Replit is a freemium option with a free daily-usage Starter plan limited to one live project and one background task at a time. Playcode has a free plan with three projects or published sites, one collaborator, and required branding, which better fits a small project-based setup. JupyterLab is free, open-source software that can be installed locally with pip. AWS Cloud9 has no additional Cloud9 charge, while AWS resource usage is billed at normal rates. Datalore offers a free cloud tier with 120 CPU S machine hours, 10 GB storage, two parallel notebooks, and one workspace. Deepnote has a forever-free plan with up to three editors and five projects. CodePen suits public front-end experiments, with unlimited public Pens on its free plan. For broader browsing, see Coding Playgrounds, Cloud Development Environments, and AI Coding Assistants.

Verdict

Choose Coder if your team wants self-hosted development environments with flexible access and a controlled route to delegate agent work. Its strongest differentiator is the combination of infrastructure control and agent execution that keeps LLM credentials out of workspaces. Look elsewhere if you want a managed, low-setup coding environment, need more than five active agents without moving to AI Premium, or require Premium capabilities such as high availability and SLA-backed support.