COSRX The 6 Peptide Skin Booster Serum: The Count Is Right, and One of the Six Is a Collagen

Product names carrying a number invite a count, and counts usually go badly. This one does not. COSRX declares forty-four ingredients, and exactly six of them are peptide-class, sitting in an unbroken run at positions seven to twelve.

That is the uninteresting part, and it had to be checked first. The interesting part is what those six turn out to be.

The six

Weights from PubChem and UniProt, counts through NCBI E-utilities, all read 2026-09-16. The copper and Argireline figures match those already published on our skin-peptide evidence page, which is the point of using a public register: the numbers reproduce.

One word covering a 247-fold range

The bottom row of that table and the row above position 9 are separated by more than two orders of magnitude. Copper tripeptide-1 is 402.9 daltons. The human collagen chain that sh-polypeptide-121 copies is 99,369.

Chemistry generally uses peptide for short amino-acid chains and protein for long ones, with the conventional line drawn near fifty residues. sh-Polypeptide-121 is on the far side of that line by a wide margin — its supplier, Geltor, calls it biodesigned vegan human type 21 collagen, made by fermentation from a sequence based on human genetic information, and the reviewed human entry for collagen alpha-1(XXI), UniProt Q96P44, runs to 957 amino acids.

INCI naming is a register of ingredient identities, not a chemical taxonomy, so a collagen and a tripeptide can sit two lines apart under the same reader-facing word. Nothing on the package distinguishes them, and the arithmetic of why that matters is on our guide to reading a peptide ingredient list.

Two of six clear the penetration threshold, which is above the category norm

The 500-dalton rule holds that compounds much above 500 Da do not passively cross intact stratum corneum in useful amounts. It is a heuristic rather than a law — our skin-peptide page covers the cyclosporin work showing heavier molecules engineered to penetrate better — but it is the number this whole category argues with.

Two ingredients here are under it: copper tripeptide-1 at 402.9 and the SYN-AKE peptide at 495.6. For comparison, the common cosmetic peptides catalogued across this site — palmitoyl tripeptide-1 at 578.8, palmitoyl pentapeptide-4 at 802.1, acetyl hexapeptide-8 at 887.0, acetyl octapeptide-3 at 1,075.2 — are all above it. Most peptide products contain none of the two exceptions, or one. This formula contains both.

That is a statement about molecular weight and nothing more. It is not evidence that anything reached a dermis, which no ingredient list can show.

Where the research is, and where it is not

The six split sharply. Copper tripeptide-1 and acetyl hexapeptide-8 carry essentially the whole file: 188 and 50 records. Both already have evidence entries here — copper peptides and Argireline — and both entries are about how much less those numbers prove than they look, since the copper literature is dominated by laboratory work and reviews written by a company selling the ingredient.

The other four total eight records between them. Palmitoyl tripeptide-8 has 5, oligopeptide-68 has 2, SYN-AKE has 1, and sh-polypeptide-121 has none under its INCI name. Geltor states that HumaColl21's efficacy is supported by in vitro and third-party clinical study results; those are not in the indexed literature, so a reader cannot open them. A supplier's unpublished data is not a reason to doubt an ingredient. It is a reason to be precise about what is available to check, which here is nothing. The one placebo-controlled trial of the SYN-AKE peptide is in a journal PubMed does not index, which is why a record count understates it; the SYN-AKE entry reads that trial.

Two ways to miscount this label

Both are worth flagging because they push in opposite directions.

Counting up. Positions sixteen to twenty-one are serine, alanine, glycine, threonine, arginine and proline — free amino acids, not peptides. They are the monomers, listed individually, common in hydrating complexes. Add them and the count becomes twelve.

Counting by prefix. Polyquaternium-51 at position thirty-eight begins with poly and is not a peptide; it is a polymer that mimics a phospholipid, with 11 PubMed records under its own and its trade name. Pattern-matching on names would include it.

The brand's six is the correct six, which after a run of labels where the number in the name did not survive a count is a finding in itself.

What six names on a label add up to

It establishes an accurate product name, six peptide-class ingredients all at 1 percent or less, two of them below the penetration threshold most of the category fails, and a size range inside one bottle that runs from a tripeptide to a collagen. It establishes that two of the six carry nearly all the published research and four carry almost none.

It does not establish what the serum does to skin. We have not used it, and a list of ingredients cannot answer that question for any product.

The same brand's four-peptide hydrogel eye patch reads very differently on the same test: its peptide count also holds up, but all four sit in the last quarter of a 41-name list, below a vitamin, a nucleoside and a stimulant with larger published files than any of them.

Sources

• COSRX, The 6 Peptide Skin Booster Serum — published ingredient declaration, ingredient listing, read 2026-09-16.
• PubChem compound records read 2026-09-16: CID 71587772, CID 71587328, CID 71465152, CID 163285842, CID 11498521.
• UniProtKB Q96P44, Collagen alpha-1(XXI) chain, Homo sapiens, reviewed — UniProt, 957 residues, 99,369 Da, read 2026-09-16.
• Geltor, HumaColl21 supplier page — Knowde, read 2026-09-16.
• PubMed record counts via NCBI E-utilities, run 2026-09-16.
• Bos JD, Meinardi MMHM. The 500 Dalton rule for the skin penetration of chemical compounds and drugs. Exp Dermatol 2000 — PubMed 10839713.