Comparisons / Comparison
Headless browser vs remote browser
Understand why display mode and deployment location are independent choices.
Headless describes how a browser runs without a visible browser window. Remote describes where the browser runs relative to its client. They answer different questions: a remote browser can be headless or headed, and a local browser can also use either mode.
Two independent choices
| Deployment and mode | Typical use |
|---|---|
| Local, headed | Interactive development and manual inspection |
| Local, headless | Automated tests on a workstation or CI worker |
| Remote, headed | A shared live browser with a viewer or human takeover |
| Remote, headless | Browser workers controlled through a network endpoint |
A viewer is a separate capability supplied by a runtime or streaming layer. The existence of CDP access alone does not establish that a service offers a viewer or permits human input.
What headless does and does not decide
Headless mode changes how the browser presents its interface. It does not choose the profile retention policy, process isolation boundary, proxy, or session timeout. Chrome's headless documentation explains the browser-side mode and command-line use.
Do not assume headless is always faster or uses a fixed amount less memory. Page content, viewport, video, extensions, active tabs, and runtime configuration can materially change resource use. Measure the workload you intend to run.
What remote does and does not decide
Remote deployment introduces a connection between the client and browser. It moves page execution and website requests into the browser environment. It does not by itself change whether the target site accepts automation, keeps a login valid, or responds quickly.
Choose the deployment based on dependency management, operational ownership, network location, and session requirements. Choose the display mode based on debugging and workflow needs. If human participation is required, test the viewer and input handoff rather than assuming that a headed process is sufficient.
Keep comparisons fair
When comparing local and remote runs, hold display mode and viewport constant where possible. Otherwise, a change in rendering behavior can be confused with a change in network or infrastructure performance. Record browser version, client location, runtime location, proxy route, and profile state.
For Remote Browser's human observation path, see live preview and recording. For deployment decisions, continue with local vs remote browser automation. Use the benchmark collection to plan measurements before making performance claims.