speedlify2

Lighthouse lab metrics and Chrome UX Report field data for a list of sites, recorded on every run and compared over time.

Measurement runs · Embed a score · Field data from the Chrome UX Report ·

mlcommons.org/en

https://mlcommons.org/en

Data from Aug 21, 2026, 4:44 AM UTC — 13h old.

Built Awesome · 1 measurement · since Aug 21, 2026 · last attempt 13h

This URL redirects to https://mlcommons.org/.

Lighthouse follows the redirect, so the numbers below describe the destination page, not this one. Not yet confirmed as a move; it needs to lead to the same place on several consecutive runs first.

Stack

Detected on the same page load as the accessibility pass — the generator from <meta name="generator"> or the shape of the page, the host from response headers.

Cloudflare Hosted by Cloudflare

Core Web Vitals field data, origin-level

Real Chrome users, 28-day window (2026-07-22 → 2026-08-18). This is what Google measures you on.

LCP 2.33 s good
INP 116 ms good
CLS 0 good

Scores

59
Performance
no change
88
Accessibility
no change
96
Best Practices
no change
92
SEO
no change

Lab metrics Lighthouse, mobile

Metric Current Previous Change Best Worst Trend
LCP Largest Contentful Paint 16.89 s — — 16.89 s 16.89 s
CLS Cumulative Layout Shift 0 — — 0 0
TBT Total Blocking Time 260 ms — — 260 ms 260 ms
FCP First Contentful Paint 3.61 s — — 3.61 s 3.61 s
Speed Index 5.08 s — — 5.08 s 5.08 s
TTFB Time to First Byte 500 ms — — 500 ms 500 ms

What LCP is made of

A flat LCP number tells you something regressed. The subparts tell you whether it was the server, resource discovery, the download, or rendering.

element render delay 1.19 s
time to first byte 50 ms
resource load duration 42 ms
resource load delay 13 ms

Field data Chrome UX Report, origin-level

Real Chrome users, 28-day trailing window (2026-07-22 → 2026-08-18). This is what Google measures you on.

Metric p75 Rating Distribution (good / needs work / poor)
LCP CWV 2.33 s good 78.4% good
INP CWV Interaction to Next Paint — field only, Lighthouse cannot measure this in a lab run 116 ms good 89.7% good
CLS CWV 0 good 95.3% good
FCP 1.46 s good 83.5% good
TTFB 975 ms needs-improvement 65.7% good

Lab vs field

Lab LCP 16.89 s · field p75 2.33 s · real users are 0.14× faster than the lab.

Field history 25 weeks

Weekly p75 from the CrUX History API, backfilled in a single request.

Week ending LCPINPCLSFCPTTFB
2026-08-15 2.34 s 114 ms 0 1.47 s 992 ms
2026-08-08 2.42 s 130 ms 0 1.52 s 1.03 s
2026-08-01 2.42 s 138 ms 0 1.50 s 1.02 s
2026-07-25 2.35 s 127 ms 0 1.43 s 984 ms
2026-07-18 2.45 s 123 ms 0 1.50 s 1.02 s
2026-07-11 2.24 s 115 ms 0 1.42 s 945 ms
2026-07-04 2.34 s 112 ms 0 1.47 s 945 ms
2026-06-27 2.31 s 112 ms 0 1.47 s 934 ms
2026-06-20 2.23 s 112 ms 0 1.42 s 920 ms
2026-06-13 2.27 s 111 ms 0 1.48 s 951 ms
2026-06-06 2.23 s 136 ms 0 1.50 s 987 ms
2026-05-30 2.38 s 127 ms 0 1.59 s 1.09 s

Page weight

Metric Current Previous Change Best Worst Trend
Page Weight 4.34 MB — — 4.34 MB 4.34 MB
Requests 94 — — 94 94
Third-party Weight 243.1 kB — — 243.1 kB 243.1 kB
Third-party CPU 96 ms — — 96 ms 96 ms
Unused JS 95.9 kB — — 95.9 kB 95.9 kB
Unused CSS 56.7 kB — — 56.7 kB 56.7 kB

By resource type

image 3.25 MB · 46 req
other 425.8 kB · 9 req
script 403.7 kB · 22 req
document 137.2 kB · 2 req
stylesheet 73.1 kB · 6 req
font 57.7 kB · 3 req
third party 772.8 kB · 14 req

Third parties 3 entities

243.1 kB transferred, 96 ms of main thread time. This is the number most likely to regress without anyone on your team shipping anything.

Google Tag Manager 156.6 kB · 83 ms CPU
Google Fonts 59.6 kB · 0 ms CPU
cloudflare.com 26.8 kB · 13 ms CPU

Page health

Metric Current Previous Change Best Worst Trend
Main Thread Work 2.41 s — — 2.41 s 2.41 s
Long Tasks 10 — — 10 10
DOM Elements 772 — — 772 772
Axe Violations Violating nodes from a full axe run — stricter than Lighthouse's subset 12 — — 12 12
A11y Failures 4 — — 4 4
A11y Failing Nodes 10 — — 10 10

Main thread work

other 925 ms
script evaluation 734 ms
style layout 338 ms
script parse compile 151 ms
parse HTML 137 ms
paint composite render 127 ms

Axe violations axe-core 4.13.0, full rule set

12 violating nodes across 6 rules, 962 passing. This is a standalone axe run, stricter than the subset Lighthouse folds into its score above — and it is the second tiebreaker in the ranking.

  • serious link-name — Links must have discernible text 5 nodes
  • critical button-name — Buttons must have discernible text 3 nodes
  • moderate heading-order — Heading levels should only increase by one 1 node
  • moderate landmark-unique — Landmarks should have a unique role or role/label/title (i.e. accessible name) combination 1 node
  • moderate region — All page content should be contained by landmarks 1 node
  • moderate skip-link — The skip-link target should exist and be focusable 1 node

Accessibility failures 4 of 30 audits

Tracked as a count, not just the score — the weighted score is coarse enough that you can ship a broken component and barely move it.

  • link-name — Links do not have a discernible name 5 nodes
  • button-name — Buttons do not have an accessible name 3 nodes
  • heading-order — Heading elements are not in a sequentially-descending order 1 node
  • skip-link — Skip links are not focusable. 1 node

Hygiene

Changes rarely — which is exactly why a change is worth noticing.

HTTPS HSTS CSP Clickjacking protection Origin isolation Trusted Types bfcache h2 1 console error

Measurement log

Every stored run for this URL, newest first. Failures are kept as records — "this site was down on Tuesday" is data too.

When PerfLCPCLSTBTWeight RunsSpreadBenchTook
Aug 21, 2026, 4:44 AM UTC 59 16.89 s 0 260 ms 4.34 MB 3/3 ±6 2,469 36.3s

Embed this score

Live below, then the markup that produces it. Drop that on any page to show mlcommons.org/en's scores, updated whenever this instance re-measures — more about the component. Don’t forget to bundle the speedlify-score.js component code into your site! 

<speedlify-score speedlify-url="https://zachleat.github.io/speedlify2/" url="https://mlcommons.org/en"></speedlify-score>