LLMS.txt Adoption Tracker

8.7% of the top 1,000 websites publish an llms.txt file, as of June 2026.

8.7%
Adoption Rate
87/1,000
Sites With a File
Monthly
Update Frequency

Last updated: 2026-06-29 · Source list: Tranco June 2026 (2026-06-23)

Adoption Over Time

We re-scan the top 1,000 every month and keep every data point, so this line only grows over time.

0%4%8%11%15% June 2025June 2026

llms.txt vs. llms-full.txt

Some sites publish the short index file (llms.txt), some add the full-content file (llms-full.txt), and a few publish both.

87
have llms.txt
15
have llms-full.txt
15
publish both files

72 publish only llms.txt · 0 publish only llms-full.txt

Adoption by Category

Where adoption is concentrated among the top 1,000, by site category.

Technology
36.4% (4/11)
Video Streaming
16.7% (1/6)
Publishing
14.3% (1/7)
Other
8.5% (10/118)
E-commerce
6.9% (2/29)
News & Media
4.8% (1/21)
Social Media
0% (0/10)
Government
0% (0/10)

Notable Adopters

High-ranking sites that already publish an llms.txt file.

cloudflare.com
Rank #3
azure.com
Rank #27
github.com
Rank #30
fastly.net
Rank #34
wordpress.org
Rank #42
digicert.com
Rank #44
adobe.com
Rank #66
workers.dev
Rank #83
opera.com
Rank #93
samsung.com
Rank #94
sentry.io
Rank #98
wordpress.com
Rank #99

Methodology

Transparent, automated, and re-run every month.

Each month we take the Tranco top 1,000 domains (a research-grade ranking that averages several traffic sources) and request both /llms.txt and /llms-full.txt over HTTPS with a clearly identified crawler. The top 1,000 and top 10,000 views are derived from the same crawl. The list used for this run was published June 2026 (2026-06-23).

A site is counted as an adopter only when it returns a real text file (HTTP 200 with plain-text content). HTML pages, empty files, and soft-404s do not count — those are recorded as “no file”. When a site can’t be reached (timeout, DNS/TLS failure, or it blocks our crawler with a 401/403/429/5xx), we record it as unknown rather than guessing. To keep unknowns to a genuine minimum we try both the apex and the www host and retry transient failures with a longer timeout before giving up. This run reached a verdict on 549 of 1,000 sites; 451 were unknown (most are infrastructure, CDN, or API domains that don’t serve a website at the root).

The headline rate is deliberately the conservative measure: confirmed adopters as a share of all sites in the sample (87 ÷ 1,000 = 8.7%) — unknowns stay in the denominator rather than being dropped. We report it this way (instead of adopters ÷ reachable sites) so the number can never be quietly inflated by counting an unreachable domain out of the sample, and so every month stays directly comparable on the same fixed base. For reference, among only the 549 sites we could reach this run, 15.8% publish an llms.txt file. Every monthly result is stored permanently, so the trend above is an immutable time series, and large month-over-month swings are flagged for human review before publishing.

Want to get this right?

Read our guide to writing an llms.txt file that AI engines actually use.

Read the LLMS.txt Best Practices Guide

Want to Implement LLMS.txt?

Use our free LLMS.txt generator to create a compliant file for your website.