Peptuvia Marketplace Transparency Dataset

Request the export

The export is the same data as a file, plus the two tables the price directory doesn't carry. We send it by hand rather than publishing a download link — tell us what you're building and we'll send the cut that fits.

• •JSON — the full release: price table, per-source fulfilment medians, COA coverage.
• •CSV — the price table, one row per kit.
• •CSV — the source table: dispatch and transit medians, COA coverage.

Sources — dispatch, transit, COA coverage

Both medians are measured from real orders: dispatch is confirmed payment to carrier hand-off, transit is hand-off to carrier-confirmed delivery. A blank means we don't yet have enough delivered orders from that source for a median to mean anything — read it as unmeasured, not as fast.

The last two columns are different kinds of evidence. COAs we publish are certificates we hold and show you at the COA library. Results the source reports come from that vendor's own third-party testing log, which we republish as numbers — the certificates themselves name the supplier, so we don't publish them. Each result covers the batch it names.

COA coverage

10 of 96 listed compounds have at least one published certificate of analysis — 14 results covering listed compounds, of which 6 carry the manufacturer lot the result covers. Where a lot is absent, the record does not identify which lot was tested.

How to cite this

Cite the version and the snapshot id. The version pins the schema; the snapshot id is a hash of the data, so a reader can re-fetch and tell whether the figures quoted are still the figures served.

Peptuvia. Peptuvia Marketplace Transparency Dataset, v1.0 (edition 2026-09-01, snapshot 1a182b321506). https://peptuvia.com/dataset. Licensed CC BY 4.0.

Methodology

scope

Every in-stock offer listed on the Peptuvia storefront at generation time, from every source whose brand appears in the public vendor directory. Sources are named by their storefront brand; upstream manufacturer names are not published. One row is one kit from one source; the row key is (vendor_slug, sku), since two sources can list the same size under the same stock code.

price per mg

kit_price_usd ÷ (mg_per_unit × units_per_kit), on the live listed price including any active discount. Published for vial forms only: dividing an oral pack's price by tablet-milligrams yields a figure that reads as comparable to an injectable's per-mg and is not, so oral rows carry null.

currency

All prices are USD and are the price a buyer pays for the kit, excluding shipping. Shipping is charged per source on a tiered schedule and depends on the rest of the cart, so it is not folded into the per-mg figure.

lead time

median_days_to_dispatch is the median days from confirmed payment to the source handing the parcel to a carrier, computed over that source's real orders. It is null until enough of those orders exist for a median to be meaningful — null means not measured yet, not fast. The sample counts behind these medians are not published: they are per-vendor order volumes.

delivery

median_days_in_transit is the median days from carrier hand-off to carrier-confirmed delivery, computed over parcels with real tracking events, under the same withholding rule. Parcels still in transit are not counted, so this measures how long deliveries took, not how many arrived.

supplier reported results

Results a source's own third-party testing log reports, republished as numbers. They are counted separately from published_coas and are never added to it: a certificate we hold and publish is different evidence from a figure a vendor reported, and only the first has a document behind it on this site. We do not publish those certificates — the lab documents name the supplier, and routing customers around our own sources is the one thing a marketplace cannot do. Each result covers the batch it names.

coa coverage

In the coverage table, published_coas counts the certificates of analysis in the public library for compounds currently listed. The per-source published_coas in the source table is counted over the WHOLE library instead, so it agrees with the vendor directory; the two therefore do not sum. share_of_compounds_with_a_published_coa is compounds with at least one published result ÷ compounds listed. coas_with_a_published_lot counts those results we hold a manufacturer lot for; where a lot is absent, the record does not identify which lot was tested. See coverage_note: this is a coverage map, not a testing claim.

exclusions

Out-of-stock offers, sources not in the public vendor directory, and product categories withheld from crawlers are absent from every table. Figures are recomputed on each release; nothing is back-filled or smoothed.

Versioning

The version pins the schema — the field names, units and meanings a citation depends on. Additive changes ship as 1.x; anything that would change what an existing field means gets a new major version, and everyone holding the current one gets told. The data itself is regenerated hourly, so cite the snapshot id: it is a hash of the tables, which means re-fetching and comparing tells you whether the figures you quoted are still the figures being served.

• v1.0 (2026-08-26) — First release: offer-level price per mg, per-vendor dispatch and transit medians measured from delivered orders, and per-compound COA coverage.