On 27 October 2026 Google released Lighthouse 13.6.0, a new version of the engine that measures how fast websites load. The release notes said it would reach PageSpeed Insights, Google's free website speed tool, within a week. By 2 November the tool was already running it.
The change is in how the tool simulates a page load. PageSpeed Insights does not measure your site on a real customer's phone. It loads the page and uses a model to estimate how long loading would take on a phone over a slower connection. The new version decides differently which files hold up the page from appearing. It also estimates differently how long it takes for the main content to show when that content is text.
Barry Pollard, who works on web performance for the Chrome team, wrote that the new estimates are more accurate and "generally result in higher scores". He said sites that lean heavily on JavaScript are more likely to see a difference. Google has not said how many points scores are expected to move.
The model still has gaps. On 23 October the Lighthouse team opened an investigation into cases where the simulation badly overestimates load times on heavy sites. In one example there, a page that actually loaded in 4.2 seconds was simulated at 20.8 seconds. That investigation is still open.
A new score for the same site
If you check your site in the tool, or get a monthly report from an agency or developer, one thing matters: a score that went up since late October may come from the version switch alone. The site did not get faster. Only the measurement changed. Comparing a test from before the switch with one from after compares two different models.
Every PageSpeed Insights report states which Lighthouse version produced it. Before crediting an improvement to work done on the site, check that both tests ran on the same version. If you pay for speed optimisation and receive a before-and-after report, ask to see the version number on both. A chart that tracks the score over time will likely show a jump where 13.6 took over, and that jump has nothing to do with the site.
Numbers from real visitors
A PageSpeed Insights report holds two kinds of data. The first comes from real visitors who opened the site in Chrome over the last 28 days. The second is the simulation, which is where the score comes from. Only the simulation was affected by the new version.
Google's own documentation for the tool says good lab data does not necessarily mean real users have a good experience. That makes the real-visitor numbers the ones to follow over time. They did not move because of this update, so any change in them reflects a real change in the site or in who visits it.
Chrome 159 arrives in December
The audit tool built into the Chrome browser, which developers use to test sites from their own computer, gets the new version only in Chrome 159. Chrome's release schedule lists that version for 8 December 2026. Until then a developer testing your site from the browser may get a different score from the one PageSpeed Insights shows, even for the exact same page. A gap like that over the next month is not a sign that something is wrong with the site.