A real virtual machine with root. That is the reason the product exists: you can run Docker, Kubernetes, systemd, custom kernels and anything else that a container cannot host. A workspace can also hold several boxes networked over a private mesh VPN, so you can model a real topology rather than a single dev container.

No. Every Engineering plan includes all three, plus EasyEnv AI. The plan tier decides how much compute, storage and AI you get, not which parts of the product you can open.

They were never really separate products: they shared the same machines and the same subscription. Workspace is now Engineering Lab, Academy is Engineering Courses, and the Claude pages are part of EasyEnv AI, which is included with Engineering. Every old link redirects to its new home.

Included on every tier, with a monthly token allowance that scales with the plan. You can use the bundled models, sign in with your own Claude account, or run an open model locally on the box when the work cannot leave the machine.

Those give you one container per repo. EasyEnv gives you real VMs, and more than one at a time on a private network, so you can run the database, the queue and the app as separate machines. It also carries the learning and assessment surface, which a dev environment on its own does not.

Environments are ephemeral by design. They stop on a TTL so you are not billed for idle machines, and you can archive a workspace to keep its definition without keeping it running. Recreating one from the same recipe takes seconds.