"llms.txt 擬似応答率" (llms.txt pseudo-response rate): 8.1% of top .jp sites on 9 November 2026, and how it is counted
The llms.txt 擬似応答率 (llms.txt pseudo-response rate) is the share of sites asked for /llms.txt that answered HTTP 200 but had no real llms.txt file, in most cases an HTML page, a "soft 404". In our weekly scan of 9 November 2026 it was 63 of 775 top .jp sites, 8.1%. In the same week 30 sites (3.9%) had a real file.
Published 15 November 2026. Numbers taken on 26 October, 2 November and 9 November 2026. "Pseudo-response rate" (擬似応答率) is our own name for this way of counting, not an industry term. llms.txt is a proposed text file at the root of a site that gives AI apps a short guide to its content; it is a proposal, not a standard.
What is the llms.txt 擬似応答率?
- Definition: the llms.txt 擬似応答率 is the share of sites asked for /llms.txt that answered HTTP 200 but had no real llms.txt.
- A soft 404 is a server answering 200 with a page for an address that has nothing at it, instead of 404. The page "works"; the thing asked for is not there.
- A real llms.txt is an answer that meets all four: HTTP 200;
Content-Typetext/plainortext/markdown; a body that is not empty and does not open with an HTML tag; and an answer that ends at /llms.txt on the site's own registrable domain. - A pseudo-response is an HTTP 200 answer that does not meet them: a
text/htmlpage, an empty body, a redirect to a home page.
How it is counted
llms.txt 擬似応答率 = sites with a pseudo-response ÷ sites we could check
- Denominator: sites asked once for /llms.txt, leaving out those whose robots.txt was served by another registrable domain and those whose /llms.txt request ended on another one: 775 to 782 sites per week.
- Numerator: among them, sites that answered HTTP 200 and not with a real llms.txt. Other statuses, such as 404, do not count.
- Population: the first 1,000 top .jp domains of the Tranco list (a public ranking of the most visited domains). Our scanner identifies itself and asks each site once, every Monday, from a Japanese IP. No site is named.
- Counted by status code alone, the week of 26 October 2026 shows 98 sites (12.6%) as having the file: about 3.4 times the 29 real files. The pseudo-response rate makes that gap visible.
The numbers (three weeks)
| Scan day | Sites checked | Pseudo-response (llms.txt 擬似応答率) | Real file | Pseudo-responses per real file |
|---|---|---|---|---|
| 2026-10-26 | 777 | 69 (8.9%) | 29 (3.7%) | 2.4 |
| 2026-11-02 | 782 | 66 (8.4%) | 30 (3.8%) | 2.2 |
| 2026-11-09 | 775 | 63 (8.1%) | 30 (3.9%) | 2.1 |
- All 69 sites of 26 October 2026 answered
text/html. - The rate slipped from 8.9% to 8.4% to 8.1%, but three readings do not make a trend.
- The real count hardly moved: 29, 30, 30.
Check your own site
- Open
your-domain/llms.txtin a browser. - Run
curl -I https://your-domain/llms.txtand read the status and theContent-Type. - If you have no file and still get 200 and
text/html, your site gives a pseudo-response. Change the server or site builder so that unknown addresses return 404. - The free check reads the file itself, not only the status: https://pw.migurei.ai/en/free-scan
Basis and limits
- "Pseudo-response rate" is our own name. It is not an industry term, and we do not know whether other surveys count this way.
- Top .jp sites only. About 780 of the first 1,000, one list, as seen from our scanner. Other sites may differ.
- We do not know why the sites answer 200. We have not looked at their servers.
- Our filter does not check syntax. A file served as plain text with wrong contents counts as real; a real file served with an odd
Content-Typecounts as a pseudo-response. - Whether any AI app reads llms.txt is a separate question; this number does not show it.
- Source: the same rows of our weekly published CSV (the "llms.txt present" and "llms.txt answered 200 with something that is not the file" rows): https://pw.migurei.ai/data/japanese-sites-and-ai-bots/2026-W46.csv . Level: observed in our lab.
- Next: the scan continues every Monday; the next numbers are those of 16 November 2026.
Related pages
- Three weeks in full, and the gap against counting by status code: llms.txt on top .jp sites: real files and HTML soft 404s
- Blog (why we raise this measure): https://migurei.ai/en/blog/llms-txt-pseudo-response-rate
- 日本語
This page is written by migurei / Parallel World about our own weekly scan. The free check is our own service.