Uncensored subscription sites number releases by publication date rather than by
label. The common shape is 010124-001 — six digits encoding a date, then a
sequence number for that day's releases. A minority of sites use plain
sequential serials instead. Neither convention includes a studio prefix, because
these platforms publish continuously rather than issuing catalogued products.
The consequence is that the number tells you when rather than who, which is the opposite of what a studio code tells you.
Why do these sites number by date instead of by label?
Because their publishing model is a daily feed, not a catalogue of discrete releases.
A studio issues numbered products into retail channels, so a serial scoped to a label is the natural key. A subscription site publishes on a schedule to existing subscribers; there is no spine to print a number on and no retailer to stock it. Numbering by publication date gives a key that is unique, sortable, and generated automatically without anyone allocating it in advance.
It also means the number is genuinely informative in a way a studio serial is
not: 010124-001 was published on 1 January 2024, first release of that day.
You cannot derive a date from SSIS-500 at all.
What format does each site use?
The layouts differ in separator and digit count, which is usually enough to tell them apart.
| Site | Format | Example | Reading |
|---|---|---|---|
| Caribbeancom | MMDDYY-nnn |
010124-001 |
Date, hyphen, daily sequence |
| Caribbeancom Premium | MMDDYY_nnn |
010124_001 |
Same, underscore separator |
| 1Pondo | MMDDYY_nnn |
010124_001 |
Date, underscore, daily sequence |
| 10musume | MMDDYY_nn |
010124_01 |
Two-digit sequence |
| Pacopacomama | MMDDYY_nnn |
010124_001 |
Date, underscore, daily sequence |
| Heyzo | HEYZO-nnnn |
HEYZO-1234 |
Site tag plus running serial |
| Muramura | MMDDYY_nnnn |
080226_1253 |
Date, underscore, running serial |
| Tokyo Hot | letter + serial | n1234 |
Series letter plus running serial |
Muramura is the one to look at twice. Its shape is identical to 1Pondo's, but the
number after the underscore is a catalogue-wide running serial rather than a
per-day counter: 080226_1253 was followed by 072426_1257, so the second half
climbs independently of the date (Muramura's own release listing, checked
2026-08-03). Reading it as a daily sequence produces a plausible-looking code
that does not exist.
Note the date order: MMDDYY, not the ISO order a non-Japanese reader might
assume. 010224 is 2 January 2024, not 1 February. Misreading this is a common
reason a date-filtered search comes back empty.
Why is a date-based code ambiguous by itself?
Because the format is shared, the same string can legitimately exist on several sites at once.
010124_001 is a valid identifier on more than one platform simultaneously —
each published something first on that day. Unlike MIDV-123, where the prefix
guarantees a single issuer, a date-based code carries no issuer information at
all. The site name is part of the identifier even though it is not part of the
string.
In practice this means:
- A bare date-based code pasted into a general search box is under-specified
- Two sources quoting "the same code" may be describing different releases
- Any database storing these codes has to store the site alongside them, or its records collide
When you record one of these, record the site with it. When you receive one without a site, treat the site as missing information rather than assuming.
How do the sequential-serial sites differ?
Sites like Heyzo and Tokyo Hot use a running serial, which behaves much more like a studio code — unique on its own, but carrying no date.
Heyzo's form is a site tag plus an incrementing number, so HEYZO-1234 is
unambiguous globally and orders releases chronologically without stating a date.
Tokyo Hot's form is a letter plus a serial, where the letter distinguishes
internal series rather than dates.
The letter-plus-digits shape is the one most often mistaken for a broken studio code, because it looks like a prefix that lost its hyphen. The tell is the digit count and the absence of a recognisable label prefix — a single letter is not a studio prefix, and no studio prefix in the reference table is one character long.
How are these different from a -U file?
Fundamentally: one is a catalogue convention, the other is a claim written into a file name.
| Site-native uncensored release | -U suffixed file |
|
|---|---|---|
| Identifier | Date-based or site serial | A studio code plus a letter |
| Issued by | The publishing platform | Whoever renamed the file |
| Catalogue record exists | Yes, on that site | No |
| Verifiable before watching | Yes, via the site listing | No |
A platform that publishes without mosaic never needs a suffix, because
everything it publishes is the same. A suffix only makes sense as a contrast
with a version that has mosaic — which is why -U appears attached to studio
codes and never to 010124-001. The suffix families are covered in
what do the suffixes -C, -U and -4K mean?.
How should you search a date-based code?
Search it on the site that published it, with the separator preserved.
Three things break these lookups:
- Swapping the separator.
010124-001and010124_001are different conventions; some indexes normalise them and many do not. - Reordering the date. Converting to
240101or2024-01-01produces a string no catalogue contains. - Dropping the leading zero.
10124_001is not a valid form.
If the code is well-formed and returns nothing, the likely causes are the same as for studio codes: the catalogue you searched does not carry the site, or the release has been removed. See JAV code not found.
What if you have a frame but no code at all?
Then identify the performer from the frame, which works regardless of numbering convention.
Date-based catalogues are exactly the case where the number is hardest to guess from context — there is no prefix to recognise and no series to browse. Matching a face is convention-independent: our index holds 241,792 faces encoded as 512-dimension ArcFace vectors, covering 2,333 named performers (our index, 2026-08 snapshot), compared at a 0.40 similarity threshold.
Boundary worth stating: our index is heavily concentrated. It spans 106 sites, but 2,246 of 2,333 representative images come from a single source, and only around a dozen of those sites carry Japanese-style codes. Subscription-site material that never circulates elsewhere is under-represented, so absence from our results is not evidence of absence from the industry.