Incogniton’s median cold-start launch was 620 ms. Octo Browser’s was 690 ms. Multilogin’s was 740 ms. A 70 ms gap sounds meaningful until you add the proxy handshake.
In the same test, the median time to first page after the proxy handshake was 1.4 seconds. The browser’s cold start is a fraction of the time a user actually waits. A residential proxy with a slow route can add more variance than the entire browser difference.
What we measured
- Cold start: process launch to a usable first page with the proxy connected.
- Time to first page: after the proxy handshake, with the target page loaded.
- RAM per profile: resident memory after the first page load.
- Stable concurrency: simultaneous profiles completing a 30-minute scripted workflow without a launch failure, crash, or proxy drop.
The methodology defines each metric. The profile-level CSV contains the per-profile launch times.
Why a pilot beats a ranking
Your proxy provider, route, target site, browser version, and hardware will change the result. A browser that is 70 ms faster on a lab rig may be slower on your route if its proxy integration adds a retry or a DNS lookup.
Run a pilot with the same proxy mix and the same target workflow. Measure the whole path, not just the cold start. If you are choosing between the top three, the feature differences — two engines, cloud phones, RPA, mobile fingerprints, Docker — will usually matter more than 70 ms.
What would be a red flag
- A vendor quoting a launch time without the hardware, proxy type, and target page.
- A benchmark that measures process launch but calls it “page load.”
- A speed claim that ignores RAM and concurrency, which determine how many profiles a machine can actually run.
The benchmark exists to make those trade-offs explicit, not to turn a 70 ms difference into a slogan.