Ask the four Faymas domains for a page that does not exist and you get four different answers: faymas.com.in returns a real 404 on 18 of 18 bogus addresses, faymas.in returns HTTP 200 for any address shaped like a known route (12 of 12) and a real 404 only when the shape is unknown (4 of 4), faymas.net returns HTTP 200 with a page containing zero words (19 of 19), and faymas-in.com returns HTTP 200 carrying its actual homepage, tagged index, follow (18 of 18). Measured on 2026-10-23 across 80 requests (76 invented addresses plus 4 real pages as controls), server-rendered HTML only.
This guide analyses published prompt text. No generation tests were run.
Why the missing page is worth measuring
A 404 is the smallest promise a site makes: this address is not a page. It is also the easiest thing to measure. You do not need a crawler budget, an API key or a keyword tool. You invent an address, you ask for it, you write down what comes back.
We did exactly that on 2026-10-23. Four domains, 19 invented addresses each, plus one real page per domain as a control. The addresses were built in four shapes: one that mimics the site's own content path, one that mimics a section path, one that matches no known route at all, and one that is a real slug with its last characters trimmed off. Every response was recorded with its HTTP status, byte size, visible word count, <title>, canonical URL and robots meta tag.
The result is not a list of broken links. It is a picture of how much of each site's address space is actually bounded.
How we tested
The sample was small on purpose. Nineteen addresses per domain is enough to tell a template from a coincidence, because a template returns the same bytes every time. It is not enough to claim a percentage of the whole site, so nothing here is stated as a share of all pages.
Two rules kept the numbers honest. First, every "is this an error page" judgement was made on visible text, not on raw HTML: the string "the page you're looking for isn't found" also appears inside the JavaScript bundle of pages that are perfectly fine, so counting it in raw HTML would have marked every page on the site as a 404. Second, each domain got a real page as a positive control, so that a zero result could not be blamed on a broken parser.
| Control page (real) | Status | Bytes | Visible words | Title present | Canonical |
|---|---|---|---|---|---|
| faymas.in/prompt/old-post-box-c7afabcb | 200 | 524,503 | 773 | yes | self |
| www.faymas.net/happyhorse-ai-video-generator | 200 | 184,362 | 707 | yes | self |
| faymas-in.com/night-bike-petrol-pump-… | 200 | 81,677 | 741 | yes | self |
| faymas.com.in/about-us/ | 200 | 193,707 | 343 | yes | self |
All four controls returned real content, a title and a self-referencing canonical. So when the invented addresses came back empty, that was the site's answer, not our failure to read it.
What each domain answers
| Domain | Status on invented addresses | Bytes | Visible words | Title | Canonical | Robots |
|---|---|---|---|---|---|---|
| faymas.com.in | 404 — 18/18 | 185,092 (all) | 54 | "Page Not Found | Faymas" | none | follow, noindex |
| faymas.in — known route shape | 200 — 12/12 | 351,080–368,226 | 52–54 | "Prompt Not Found | Faymas" / "Category | Faymas" | https://faymas.in | noindex |
| faymas.in — unknown route shape | 404 — 4/4 | 390,706–390,785 | 19 | "Faymas" | https://faymas.in | noindex |
| www.faymas.net | 200 — 19/19 | 99,357–100,779 | 0 | none | none | noindex |
| faymas-in.com | 200 — 18/18 | 66,781 (all) | 95 | "faymas in -" | https://faymas-in.com/ | index, follow |
faymas.com.in: the one that says no
Every invented address got a real 404. Not a soft one: the HTTP status line said 404, the title said "Page Not Found | Faymas", and the robots meta tag said follow, noindex. All 18 responses were the same 185,092 bytes, which is what you expect from one error template.
It also did something none of the others did. When we asked for /about-u/ — the real page /about-us/ with one character removed — it answered with the real page: status 200, canonical https://faymas.com.in/about-us/, three times out of three. That is WordPress guessing the intended URL. It is a small courtesy, and it was the only one on offer.
faymas.in: the status code depends on the shape of the address
This is the most interesting result of the day. faymas.in does not have one answer for a missing page. It has two, and which one you get depends on whether your address matches a route the application knows about.
Ask for /prompt/walnut-7a5cf75 or /category/harbour-ea55c3d1 — shapes the app has handlers for — and you get HTTP 200 with a rendered "Prompt Not Found" or "Category" page, 52–54 visible words, and a canonical pointing at the homepage https://faymas.in. The visible text was byte-identical across all twelve requests (one md5, twelve hits), so this is one template, not twelve near-misses.
Ask for /zz/compass/orchid-710c8e3f — a shape with no handler — and you get a real 404, nineteen words, and the plain-language line "Oops! The page you're looking for isn't found".
Two things follow. First, the site is not lying to crawlers about those 200s: every one of them carries noindex and a canonical that points home. Second, /prompt/<anything> is a permanently open door. It returns a page for any string you put after it. That matters because the 8-character hash at the end of every real slug is part of the address: we trimmed it (/prompt/old-post-box) and got the not-found page back, three times out of three.
faymas.net: 200, and nothing in it
Nineteen invented addresses. Nineteen HTTP 200s. Zero visible words in every one — not "a short page", but the md5 of an empty string. No <title>. No canonical. Two <div> elements and no <h1>. The byte size varied only between 99,357 and 100,779, which is build-ID noise, not content.
There is one mitigation and it is worth stating: those empty pages are marked noindex, while the real control page carries no robots restriction at all. So the site does distinguish "this address has no page" from "this address has a page" — it just says it in a meta tag instead of in the status code.
faymas-in.com: 200, and here is our homepage
Eighteen invented addresses, eighteen HTTP 200s, all exactly 66,781 bytes. The visible text of every one of them was identical to the visible text of its own homepage — same md5 (e48adb7e) as https://faymas-in.com/. The canonical pointed at the homepage and the robots tag read index, follow.
This is the only domain of the four where an address that does not exist returns content that is not marked noindex. A canonical pointing at the homepage is a hint to consolidate, not an instruction to drop the URL, and we did not and cannot test how any particular crawler reads it. What we can say, from the response itself: the status says success, the robots tag says index it, and the body is a real page with 95 words of navigation and post listings on it.
The near-miss test, side by side
| Domain | Address asked for | Status | Canonical returned | What came back |
|---|---|---|---|---|
| faymas.com.in | /about-u/ | 200 | /about-us/ | the real page (3/3) |
| faymas.in | /prompt/old-post-box | 200 | https://faymas.in | "Prompt Not Found" (3/3) |
| www.faymas.net | /happyhorse-ai-video-generato | 200 | none | empty shell (3/3) |
| faymas-in.com | /night-bike-…-for-boys/ | 200 | / (home) | homepage content (3/3) |
One character removed from a real address, four domains, four policies. Only one of them tried to work out what you meant.
What this says about the size of these sites
People ask how many pages a prompt site has, and the honest answer for three of these four domains is: more than any crawl can count, because the crawl never hits an edge.
faymas.in will render a page for any string you place after /prompt/. faymas.net will return a 200 for any path at all. faymas-in.com will hand you its homepage at any address. In all three cases the set of addresses that produce a response is open-ended; the only thing that closes it is the sitemap, which is a file the site writes about itself.
Today's sitemap from faymas.in lists 21,104 URLs, of which 21,044 are /prompt/ pages. Our measurement on 2026-10-22 recorded 20,643 /prompt/ URLs — about 400 more in a day, with nothing deleted. That number is real and useful, but it is a number the site publishes, not a number you can arrive at by walking the site. Walking it, you never get told to stop.
There is a twist here that we did not expect. The biggest site in the group is the most careful about its own missing pages: faymas.in marks every soft 404 noindex and points the canonical home. The smallest one, a four-page WordPress site, is the only one that answers an invented address with indexable content. Scale and hygiene did not correlate in this sample.
What we did not measure
We read server-rendered HTML with curl. We did not run JavaScript, so anything a site renders only in the browser after load is invisible to these numbers — which matters most for faymas.net, whose empty 200 could well become a real error page once the client-side router runs. We did not test how any search engine treats these responses; a canonical tag is a suggestion and an indexing outcome is not something we can observe from here. And 19 addresses per domain tells you the template, not the rate: none of the counts above should be read as a percentage of all pages on these sites.
FAQ
Does faymas.in return a 404 for a missing prompt page?
It depends on the shape of the address. Any address shaped like /prompt/<something> or /category/<something> returns HTTP 200 with a "Prompt Not Found" or "Category" page (12 of 12 tests, all marked noindex with a canonical pointing at the homepage); addresses matching no known route return a real 404 (4 of 4).
Which Faymas domain handles missing pages best?
faymas.com.in. All 18 invented addresses returned HTTP 404 with the title "Page Not Found | Faymas" and a follow, noindex robots tag, and it was the only domain of the four that used the 404 status code for every request we made.
Does faymas.net have a 404 page?
Not in the server response. All 19 invented addresses returned HTTP 200 with zero visible words, no title tag and no canonical; the pages are marked noindex, but the status code says the request succeeded.
Does faymas-in.com serve its homepage at addresses that do not exist?
Yes. All 18 invented addresses returned HTTP 200 with visible text identical to its homepage (md5 e48adb7e), a canonical pointing at https://faymas-in.com/ and a robots tag reading index, follow — the only domain in this test that returned non-noindex content for a missing address.
Do these sites correct a nearly right URL?
Only faymas.com.in. Asking for /about-u/ returned the real /about-us/ page with a 200 and a canonical of /about-us/ in three of three attempts; faymas.in answered a trimmed prompt slug with its not-found page, and faymas.net and faymas-in.com answered with their usual empty shell and homepage respectively.