Boxes.dev vs Continuum

At a glance

As of October 2026.

Who runs the computer

Continuum runs agents on computers you provide, and boxes.dev runs the computers for you.

Continuum's Mac app runs sessions on your Mac by default. To keep agents working after your laptop closes, you can enroll another Linux or Mac machine that you can reach over SSH or Tailscale, such as an old laptop, an EC2 instance, or a Hetzner server. Enrolling a machine installs Continuum's agent, Node, and the agent CLIs, and its sessions work in that machine's copy of your repository, with whatever databases and services you've set up there. You keep each machine secure, updated, and running. You can also connect a Cloudflare or Vercel account, and Continuum will lease short-lived sandboxes there for sessions, with an idle stop and a cost cap you set. Accounts that include Continuum Cloud can use managed runners that clone the repository, run the session, and stop.

Boxes.dev first sets up your full development environment on a cloud machine called the Template box, with the tools, databases, services, and data your app needs, so an agent can run the app and test its changes end to end. A setup agent can port the environment from your laptop, including the files, local data, agent settings, and skills you choose to upload, or build it from a GitHub repository. Boxes.dev then saves the Template box as a snapshot, and every task runs on its own devbox, a full, isolated copy of that machine. A maintenance script can keep the Template box current as your code changes, running on a schedule or when files such as a lockfile change, and each successful run saves a new snapshot. Team Templates let an admin prepare one starting environment for everyone on the team.

You'd never let an engineer ship code they hadn't run, and a full environment holds the agent to the same standard. On a boxes.dev devbox, the agent runs your whole app, with its databases and services, and tests its changes in the browser as it builds. It reproduces bugs in the real app to find the root cause instead of guessing from the code, so it writes better code and makes fewer mistakes. It can also take screenshots and record video for you to review, so you can hand it most repetitive manual testing.

Worktrees on one host, or a machine for each task

Continuum separates parallel sessions with Git worktrees, and boxes.dev gives each task its own machine.

Each Continuum session gets a managed worktree, a second checkout of the repository on its own branch, which Continuum creates and cleans up when the session ends. Sessions on the same host share its CPU, memory, services, databases, and ports, so sessions that each need their own database or a dev server on the same port need more setup, such as containers or another host. Continuum suggests spreading sessions across machines, so several agents don't compete for one laptop.

Each boxes.dev devbox is a separate machine with its own copy of your databases and services, so every task can run the whole app on its usual ports. A devbox can also hold several threads, which share its files and running services.

Reaching the running app

Both let you comment on the running app for the agent, and boxes.dev also lets you share the app and reach it from your laptop.

Continuum's Browser Preview opens the running app in a pane of the app. When you Cmd-click an element and leave a comment, Continuum adds it to the agent's conversation with the element's selector, so the next turn can change that element.

Each boxes.dev devbox runs a browser that you and the agent share, so you can watch the agent test the app, click through it yourself, or comment on page elements to show the agent what to change. Ports forwards the app to the same localhost port on your laptop and gives each devbox stable .localhost addresses. A public link shares a devbox's running app with anyone, and a Project URL gives your project one stable public address that you can point at any devbox, which suits OAuth callbacks and webhooks.

When you close your laptop

Both keep working after you close your laptop when the work runs somewhere else, and boxes.dev also sleeps machines that nobody is using.

Sessions on an enrolled machine keep running after you close your laptop, and you can follow them from the web app or your iPhone. Sessions on the laptop itself stop with it. A cloud VM used as an enrolled machine runs, and bills, until you stop it, and Continuum's sandboxes stop when the work finishes or reaches the idle stop or cost cap you set.

A boxes.dev devbox stays awake while its agent works and sleeps when nobody is using it. Sleep keeps the machine's files and memory, like closing a laptop's lid, so dev servers, databases, and terminal sessions resume where they left off, and a sleeping devbox doesn't use box-hours. The devbox stays yours until you delete it.

What comes with every boxes.dev devbox

The boxes.dev desktop app opens the same set of tools for whichever devbox you select:

• Terminal, with tabs that keep running after you close the app.
• Files, a file browser and editor with search across the workspace.
• Git/Review, for branch history and diffs, where comments on changed lines go to the agent together as one message.
• Browser, the browser on the devbox that you and the agent share.
• Ports, for the forwarding, .localhost addresses, and public links described above.
• Activity, which shows the processes using the machine's CPU and memory.

Mentioning @boxes.dev in a Slack thread starts Codex or Claude Code on a new devbox, and later mentions from you or teammates you allow send follow-ups to the same conversation, which run with your access. Work can also start from an assigned Linear issue, a schedule, or an API call. Agents can start other agents on new or existing devboxes, message each other, read earlier threads, and leave notes in a shared /project folder. You can share a thread's link with your team, let teammates read threads, and link commits to the threads that made them. The iOS and Android apps can reply to agents, browse files, and open a shell on a devbox.

Where Continuum shines

• A free app for many agents. Claude Code, Codex, Cursor, Grok, OpenCode, Antigravity, and Z.ai run through their own CLIs and logins, and an OpenRouter key adds 100+ more models.
• Several accounts side by side. You can connect more than one Claude or Codex account, each with its own usage gauge, and move a session to another account when one runs low.
• One message to several models. Broadcast sends the same message to several agents at once, shows each answer with its cost, and continues the conversation with the answer you star.
• Spending by repository and model. Usage analytics break down spending and tokens by repository, provider, model, and day.

Pricing

As of October 2026.

A box-hour is one default devbox awake for an hour, and sleeping devboxes don't use any. On Starter, 40 box-hours works out to about two hours of devbox time each working day, and Pro's 250 to about twelve, spread across as many as ten devboxes at once. Your Claude or ChatGPT plan covers the agents' usage. See Plans, seats, and box-hours.

Continuum's app is free with your own plans and keys. Plus, Max, and Ultra buy Continuum-hosted inference with a weekly budget of $25 to $1,000, and a Team subscription adds model policy, weekly caps, and one bill, priced by member. See Continuum's pricing.

Using both

Continuum and Boxes.dev can work on the same repositories, with the same Claude and ChatGPT logins. You can keep Continuum for comparing models and managing several subscriptions, and use boxes.dev when the agent should test its work in its own copy of your full running app, or when work should start from Slack or Linear.

Frequently asked questions

Is Continuum free?

Yes. The app is free with your own plans and API keys, and Continuum charges only for its optional hosted inference. You pay for any machines or cloud accounts your sessions run on.

How does Continuum on a cloud VM compare with boxes.dev?

A well-prepared VM enrolled in Continuum gives you agents that keep working after you close your laptop and that you can follow from the web app or your iPhone. You still provision, secure, update, and pay for the VM, and parallel sessions share it. Boxes.dev gives each task a full, isolated copy of your Template box, sleeps idle machines while keeping their memory, and adds a browser you share with the agent, Slack, Linear, schedules, and public URLs.

Can I use my Claude and ChatGPT plans with both?

Yes. Continuum runs Claude Code and Codex through their own CLIs and logins on each host, and boxes.dev runs them with your Claude or ChatGPT plan or an API key.

Sources

• Continuum: home, devices, Continuum Cloud, Code workbench, plan and review, sessions, accounts, providers, mobile, share, organizations, download, and pricing.
• Boxes.dev: How boxes.dev works, Update your Template box and snapshots, Startup, teardown, and maintenance, Sleep, wake, and recover a devbox, Review a page with your agent, Browse, edit, and transfer files, Preview your app, Project URLs, Commit, push, and open a pull request, Create and publish a Team Template, Connect Codex and Claude, Install and use the CLI, Work from Slack, Create an automation, Shared project storage, Organize and share threads, Use boxes.dev on your phone, and Plans, seats, and box-hours.