Open VS Code, click the account icon, flip on “Turn on Remote Tunnel Access”, and the desktop under your desk becomes a machine anything with a browser can reach. That single checkbox is the reason the best remote code editing apps have gone from niche to default in the last two years. The workstation stays where the RAM, GPU, and Docker daemon live. The keyboard moves to a Chromebook on a train, a borrowed laptop at a client site, or an iPad at a coffee shop. Nothing gets copied down, nothing gets lost when the battery dies, and the extensions, git identity, and open tabs are the same session you left an hour ago. Below are eight tools that solve this problem in slightly different ways, from a free checkbox to a self-hosted platform for a fifty-person team.
What to look for in a remote code editor
A few things separate a weekend curiosity from something you can lean on for a year.
- Session persistence. Does closing the browser tab kill your long-running build, or does the workspace keep going and reattach on the next connect?
- Port forwarding. A dev server on port 3000 on the remote box needs to reach
localhost:3000in your browser, and it should be one click, not a manual SSH incantation. - Extension parity. Marketplace extensions, LSP servers, and language toolchains should install the same way they do on desktop. Some tools swap in Open VSX, which has a smaller catalog.
- Latency handling. Local echo, deferred syntax highlighting, and predictive cursors turn a 120ms link from painful to fine. Ask before you commit to a tool over a transatlantic hop.
- Auth model. SSO, hardware keys, and per-workspace tokens matter the second more than one person uses the system. A shared password is not an auth model.
- Self-hosting vs SaaS. Self-hosted means you own the data and the bill is your hardware. SaaS means someone else patches the base image and you pay by the minute.
- Cost per seat. Cloud dev environments are usually billed per active hour. Multiply your working week honestly before comparing to a flat per-user license.
Quick comparison
| App | Best for | Host model | Free plan | Starting price | Extensions |
|---|---|---|---|---|---|
| Coder | Self-hosted team platform | Your infra | Open-source core | ~35 USD/user/mo (Premium) | Open VSX + VS Code Marketplace |
| Gitpod | Cloud dev envs, PR previews | SaaS or self-hosted | 50 hours/mo | ~15 USD/user/mo | Open VSX |
| GitHub Codespaces | GitHub-native cloud dev | SaaS | 60-120 core-hours/mo (personal) | ~0.18 USD/hour (2-core) | VS Code Marketplace |
| VS Code Tunnels | Solo dev, own desktop | Your desktop | Fully free | Free | VS Code Marketplace |
| JetBrains Gateway | JetBrains users on remote hosts | Your infra | Free client | Requires IDE license | JetBrains Marketplace |
| code-server | Self-hosted browser VS Code | Your infra | Fully free | Free | Open VSX |
| Zed | Real-time pair programming | Your infra / SSH | Free tier | ~20 USD/user/mo (team) | Zed extensions |
| Cursor | AI-first editing over SSH | Your infra | Hobby tier | ~20 USD/user/mo | VS Code Marketplace |
The apps
1. Coder
Coder is the closest thing to an internal developer platform for remote workspaces. Admins define a workspace as Terraform, and every developer gets a reproducible environment on Kubernetes, a VM pool, or bare-metal boxes in a rack somewhere. Users pick their editor (VS Code, JetBrains, or the browser) and Coder handles provisioning, idle shutdown, and identity.
Weaknesses: the setup is real work. If your team is under five people and you do not already run Kubernetes, the operational overhead outweighs the win.
Pricing: the open-source core is free; Premium adds RBAC, audit logs, and support at roughly 35 USD per user per month.
Platforms: browser workspace + native VS Code/JetBrains clients on Windows, macOS, and Linux.
Download: Coder
Bottom line: pick Coder when you want every engineer on the same reproducible image and you have someone who enjoys Terraform.
2. Gitpod
Gitpod pioneered the “one workspace per pull request” pattern. Push a branch, click a badge, and 30 seconds later a fresh container is booting with your repo cloned, dependencies installed, and the dev server warm. The .gitpod.yml file in the repo describes the environment, so onboarding a new hire is git clone plus a browser tab.
Weaknesses: the free tier is generous but not unlimited, and the SaaS runs in the EU/US, which matters for regulated codebases. Gitpod Flex (self-hosted) closes that gap.
Pricing: 50 hours per month free, then roughly 15 USD per user per month for the professional tier.
Platforms: browser, plus a desktop bridge for local VS Code and JetBrains IDEs.
Download: Gitpod
Bottom line: the fastest way to turn a repo into a click-to-code environment without owning any infrastructure.
3. GitHub Codespaces
Codespaces is Gitpod’s answer from inside GitHub. A .devcontainer/devcontainer.json file, a click on any repo, and a cloud VM boots with your code checked out and forwarded to a browser or a native VS Code window. Prebuilds cache image layers so a two-minute cold start becomes 15 seconds.
Weaknesses: it is priced by the core-hour, and heavier machines add up quickly for professional accounts. Personal accounts get a healthy free bucket that resets monthly.
Pricing: personal accounts get 60 core-hours per month free (120 on Pro), then roughly 0.18 USD per hour for a 2-core box, more for bigger machines and storage.
Platforms: browser, native VS Code, JetBrains via Gateway.
Download: GitHub Codespaces
Bottom line: if the code already lives on GitHub, this is the shortest path to a working cloud IDE.
4. VS Code Tunnels
The zero-cost, zero-infra option. Install VS Code on your desktop, sign in with GitHub or Microsoft, enable “Remote Tunnels”, and the machine is reachable at vscode.dev/tunnel/<name> from any browser. The connection is a WebSocket relayed by Microsoft, so it works from networks that block inbound SSH.
Weaknesses: single-user by design, no policy controls, and the desktop must be awake. It is a productivity tool for one person, not a team platform.
Pricing: free.
Platforms: browser (vscode.dev), plus native VS Code on Windows, macOS, and Linux as the client.
Download: VS Code Tunnels
Bottom line: the best free option for a developer with a home workstation and a laptop that struggles to compile.
5. JetBrains Gateway
Gateway is a thin launcher that connects to a remote host and runs a full IntelliJ, PyCharm, WebStorm, or Rider backend there, streaming the UI to your machine. The experience is closer to a native IDE than a browser tab because it is one, driven over a low-latency protocol.
Weaknesses: needs a JetBrains license on the backend, and the first-connect indexing can chew CPU. Not a good fit for a Chromebook with no local install.
Pricing: the Gateway client is free; the backend needs a standard JetBrains subscription (roughly 16 USD per user per month for a single product, more for the All Products Pack).
Platforms: Gateway client on Windows, macOS, and Linux; backend runs on any Linux host with the IDE installed.
Download: JetBrains Gateway
Bottom line: the only remote option that keeps the full JetBrains refactoring toolset intact.
6. code-server
Coder’s other project: a plain HTTP server that runs VS Code in a browser. Point it at a directory, put it behind a reverse proxy with TLS, and any tab is your editor. Because it uses Open VSX rather than the VS Code Marketplace, a handful of proprietary extensions (Pylance, Remote-SSH from Microsoft) are missing.
Weaknesses: you own the TLS, the auth, and the reboots. Single-user unless you script per-workspace containers.
Pricing: free and open source.
Platforms: browser client, backend runs on any Linux/macOS host (Docker image available).
Download: code-server
Bottom line: the classic self-hosted answer when you want VS Code on a Raspberry Pi, a homelab NAS, or a cheap cloud VM.
7. Zed
Zed’s collaboration story is the interesting bit for remote work. Two people can share a project, edit the same file at the same time with visible cursors, and voice-chat through the editor. Remote development over SSH turns Zed into a client for a workspace living on another box while keeping the low-latency feel Zed is known for.
Weaknesses: the extension ecosystem is younger than VS Code’s, and Windows support is still stabilizing.
Pricing: free for individual use; team plan runs about 20 USD per user per month.
Platforms: native clients on macOS and Linux (Windows in preview); remote sessions over SSH.
Download: Zed
Bottom line: the pick when pair programming and typing speed matter more than raw plugin count.
8. Cursor
Cursor is a VS Code fork with an AI agent glued to every keystroke. The remote SSH support inherits from VS Code, so a Cursor client on a laptop can drive a workstation the same way. What you gain over vanilla Remote-SSH is the agent’s ability to read across a large codebase, propose multi-file edits, and run commands in an integrated terminal that stays on the remote host.
Weaknesses: the model calls run against Cursor’s backends, so air-gapped setups need the enterprise tier with self-hosted inference.
Pricing: hobby tier free with limits; Pro is roughly 20 USD per user per month.
Platforms: native clients on Windows, macOS, and Linux; connects to any Linux SSH target.
Download: Cursor
Bottom line: if the AI workflow is doing the heavy lifting and the remote host is just where the code lives, Cursor keeps both together.
How to pick the right one
Working alone with a spare desktop under a table: turn on VS Code Tunnels. Zero cost, one checkbox, and the browser client at vscode.dev is enough for most weeks. Fall back to native VS Code when a heavier task needs it.
Small team that wants every environment identical and every laptop stateless: run Coder on your own cluster. The upfront work pays back the first time a new hire is productive on day one instead of day five.
Team that lives inside GitHub and does not want to run infrastructure: GitHub Codespaces. The devcontainer spec is the same one Coder and Gitpod read, so nothing is locked in.
Team that prefers Gitpod’s PR-preview model or needs a non-GitHub source: Gitpod.
Shop that uses IntelliJ, PyCharm, or Rider daily: JetBrains Gateway against a Linux host, since a browser VS Code will not replace ReSharper.
Hobbyist with a homelab and no budget: code-server behind Caddy or Nginx.
Pair-programming heavy team on macOS: Zed for the collaboration, with SSH remotes for the compute.
Individual developer whose main leverage is AI assistance across a large codebase: Cursor with Remote-SSH.
FAQ
Is code-server the same as VS Code? It runs the open-source VS Code build in a browser and looks nearly identical. The gap is the extension marketplace: code-server uses Open VSX, so Microsoft-owned extensions like Pylance, the C++ toolset, and the official Remote extensions are not available. Community forks cover most gaps but not all.
Is Codespaces expensive for hobby use? Personal GitHub accounts get 60 core-hours per month free (120 on Pro), which covers a couple of hours a day on a 2-core machine. Bumping up to a 4- or 8-core box burns the free bucket faster, and after that it is billed by the hour. For a weekend project on a 2-core prebuild, the free tier is usually enough.
Do tunnels punch through corporate firewalls? VS Code Tunnels and JetBrains Gateway both talk to a relay over standard HTTPS, so they work from most networks that allow web browsing. Coder and code-server run on your infrastructure, so you need whatever inbound path your firewall permits, usually a reverse proxy on 443. Strict networks that inspect TLS may still break tunnel-style connections.
Does Coder replace Docker? No. Coder orchestrates workspaces that almost always run as containers, so Docker (or containerd, or Kubernetes) is the layer underneath. What Coder replaces is the ad-hoc README that says “run this docker-compose file, then export these env vars, then hope”.
Can Gateway replace JetBrains desktop? For most workflows, yes. The remote backend is the same IDE, the same indexing, the same refactoring. The two edge cases are heavy GUI plugins that expect a local windowing system and offline work on a plane, both of which still favor a full desktop install.
