Browser fleets — a managed Selenium Grid & Playwright endpoint

You already have a working Selenium or Playwright suite. Browser fleets let you keep every line of it and swap one URL — MaxoPerf runs the browsers, and hands you a single central control endpoint that speaks both the W3C WebDriver protocol (Selenium Grid-compatible) and Playwright connect().

A fleet is a set of real chrome, firefox, and edge browsers you request by count and location. Point your existing client at the fleet endpoint and MaxoPerf allocates the browsers, records video, steps, HAR, and console for every session, and bills them as browser virtual users on your existing plan.

The one-URL swap

The only change to your suite is the server URL. Your test code, selectors, waits, and assertions stay exactly as they are — you are just running against MaxoPerf-hosted browsers instead of your own grid.

Why teams move to browser fleets

• A Selenium Grid alternative you don’t operate. No hub, no nodes, no Docker images, no autoscaling to babysit. Request browsers, connect, done.
• A Selenoid / Moon replacement. Selenoid is archived. Fleets give you the same facade-over-drivers model as a managed service, with observability built in.
• A Microsoft Playwright Testing replacement. MPT is being retired. Fleets speak Playwright connect() per engine, plus a version-agnostic CDP escape hatch.
• Full observability, no script edits. Every session gets a clean video, a synced step timeline, a network HAR, and the browser console — the same capture as a MaxoPerf browser test.
• Agent-native. Create and scale fleets from the MCP server, Loadrigo, your IDE, the REST API, or the console.

How a fleet works

1. Request browsers. Tell MaxoPerf which browsers you want, how many of each, and in which location — from the console, the REST API, the MCP server, or an AI agent.

2. MaxoPerf allocates runners. Each fleet runs on the same runner image a browser test uses. Browsers connect outbound to a gateway over a secure tunnel — there are no inbound ports and no runner addresses to manage.

3. Connect your client. You get one central endpoint. Selenium clients use the WebDriver URL; Playwright clients use the per-engine WebSocket URLs; anything Chromium can use the CDP escape hatch.

4. Watch and inspect. Live sessions appear as video tiles on the fleet’s run in the console, each with its own steps, HAR, and console.

Browser fleets vs. browser tests — when to use which

Both features run real chrome, firefox, and edge browsers on the same runner image, capture the same video/steps/HAR/console, and bill from the same VU-hours pool. The difference is who drives the browser:

	Browser test	Browser fleet
Who drives the browser	MaxoPerf runs your recorded or scripted scenario	Your own Selenium/Playwright client, running wherever your suite already runs
Best for	Load and functional scenarios authored in MaxoPerf (builder, VarioTest, or uploaded script)	An existing Selenium/Playwright suite (CI, local dev, another framework) that needs somewhere to run browsers
Lifecycle	MaxoPerf starts, drives, and ends every session	You open and close sessions against a long-lived fleet endpoint over its lifetime
Concurrency model	Set once at test create (virtual users)	Request N browsers up front, or scale a live fleet up as your suite needs more

Tip

Not sure which one you need? If your test code already exists as a Selenium or Playwright suite, reach for a fleet — it’s a one-URL swap. If you’re authoring a new scenario inside MaxoPerf (builder or script upload), use a browser test instead.

The endpoints

Every fleet exposes one central control endpoint, path-routed on the app domain — no new host, no extra DNS:

Protocol	Endpoint	Browsers
WebDriver (Selenium Grid)	https://app.maxoperf.com/browser-fleets/<fleetId>/wd/hub	chrome · firefox · edge (pick via the browserName capability)
Playwright (per engine)	wss://app.maxoperf.com/browser-fleets/<fleetId>/playwright/{chromium,firefox,msedge}	one WebSocket URL per engine
CDP (escape hatch)	wss://app.maxoperf.com/browser-fleets/<fleetId>/cdp	Chromium, version-agnostic

Note

Staging and local (Tilt) environments expose the same paths on their env-equivalent app host. The scheme is identical everywhere: https://<app-host>/browser-fleets/....

Authentication in one line

Every credential position accepts either a fleet token (mpft_…) or an account API key (mpak_…) with access to the fleet. Drop it into whichever slot your client uses:

• Selenium — basic-auth in the URL: https://<fleetId>:<token>@app.maxoperf.com/browser-fleets/<fleetId>/wd/hub
• Playwright / CDP — the ?token=<token> query parameter on the WebSocket URL
• Raw HTTP — an Authorization header, or the maxoperf:options.token capability on WebDriver

Retrieve or rotate a fleet token any time the fleet is live with GET /v1/fleets/<fleetId>/token, or reveal it on the fleet’s detail page in the console.

What gets captured

Capture is honest about what each browser and protocol supports today. WebDriver on chrome and edge is full-fidelity; Playwright captures video for every engine, with console and HAR best-effort.

Browser	Protocol	Video	Steps	Console	HAR
chrome	WebDriver	✅	✅	✅	✅
edge	WebDriver	✅	✅	✅	✅
firefox	WebDriver	✅	✅	⚠️	⚠️
chrome	Playwright	✅	❌	⚠️	⚠️
edge	Playwright	✅	❌	⚠️	⚠️
firefox	Playwright	✅	❌	❌	❌

Legend: ✅ full · ⚠️ best-effort in v1 · ❌ not available in v1.

Note

Video is always captured for every session, on every browser and protocol. Steps come from the WebDriver proxy, so Playwright sessions (which speak their own wire protocol) don’t produce a step timeline. Firefox console/HAR uses WebDriver BiDi rather than CDP and is a v1 follow-up. Playwright chrome/edge console and HAR are best-effort via CDP.

Next steps

Connect a client Endpoints, auth, and a protocol-agnostic walkthrough for wiring any Selenium or Playwright client to a fleet.

Playwright quickstart Connect a Playwright suite with per-engine projects — plus version matching and the CDP escape hatch.

Selenium quickstart Python, Java, and JavaScript clients — and how to create a fleet from the REST API.

Create fleets from AI agents Spin up and scale fleets from the MCP server, your IDE, or Loadrigo.

Billing & limits How fleets are metered, plus session, idle, and lifetime timeouts and error behavior.

See also

• Browser tests — author a new scenario inside MaxoPerf instead of driving your own client.