Public peptide test results: where to find them and how to weigh what you read

Where public peptide test results live

Public results come from three kinds of source. They answer different questions, so it helps to know which one you are reading.

Source	What it holds	Who chose the sample	What you can check
Finnrick	10,405 published tests covering 352 sellers, to 20 Sep 2026	Buyers donate vials; samples are anonymised before the lab; seller-submitted samples are labelled	Pass rates by seller and peptide; original lab PDFs for paying members
Peptigrity	13,189 tests from 714 shops; a directory of 32 labs	Whoever ordered the test, often the seller; some COAs are imported from lab portals; sellers’ in-house tests are not accepted	Test counts, average purity and a verifier flag per lab; a lab-verified status per COA
PeptideBenchmark	A ledger of 14,494 COA records	Whoever ordered the test, often the seller	Reused or duplicate records, and whether lab lookup links still respond
Disclosed Labs	6,401 COAs; a credibility index of 40 labs	Whoever ordered the test, often the seller	4,179 COAs checked against lab portals; purity spread and safety-test reporting per lab
Merit Verified	6,756 COAs on 1,166 lots from 166 sellers	Whoever ordered the test, often the seller	A score per seller and a quality tier per lab
PeptIQ	A community COA database inside a peptide app	Not stated	Trust scores for 46 labs, from document patterns; the formula is not published
Lab lookups	Janoshik’s public database, Freedom Diagnostics’ search by company, Vanguard’s reporting hub	Whoever ordered the test	The lab’s own copy of the report
BatchGuild	590 COA listings, 306 of them backed by a report	Whoever ordered the test	Whether a listing is backed by a report or only by a source link

Most of these sites collect certificates; they do not test vials. Finnrick, VialAudit and ChemVerify are the exceptions: they buy or receive product and send it to labs. VialAudit’s partner labs are not named; ChemVerify uses Janoshik, MZ Biolabs and Vanguard. Lab lookups usually show only what a client chose to publish. Janoshik lists a report publicly on request, and Vanguard’s hub shows only reports the client made public.

Business models differ too. Peptigrity says it takes no ads, affiliates or paid placement. PeptideBenchmark sells sponsored placements, and Disclosed Labs sells placements to sellers it grades A or B, both stated on their own pages. Our comparison of rating sites covers each one.

How Finnrick results work

Finnrick classes each test as pass, fail or incomplete under a standard you choose. Two standards are public, and three more are for members.

Standard	Purity	Fill vs label	Other
Purity Focus (default)	At least 99.5%	Within ±35%	None
Finnrick Standard	At least 99.5%	Within ±25%	Label score at least 0.5
Accuracy Focus (members)	At least 98%	Within ±20%	Label score at least 0.5; fails endotoxin above 40 EU per sample, where a result exists

Overfill and underfill both fail. The label score rewards a batch ID printed on the vial. Each test’s weight halves about every two years, so a 2025 test counts for less than a 2026 one. Sellers with few tests are pulled towards the average for that peptide. Finnrick dropped its A to E letter grades on 6 July 2026, so a screenshot with a letter grade comes from the old system.

Totals by product show how wide the range is. Under Purity Focus, Finnrick’s data to 20 September 2026 reads:

Product	Tests	Pass	Fail
Retatrutide	3,858	2,988	824
GHK-Cu	807	695	104
BPC-157	791	391	395
Tesamorelin	604	274	326

Pass and fail do not add up to the total, because some tests are incomplete.

Five checks before you trust a public result

How many tests, and how recent

One test describes one vial. A seller’s record of 40 tests from 2025 says little about a batch made this month. Look for the count, the dates and the lot numbers, not only the headline percentage.

Which lab measured it

Finnrick publishes a pass rate for each partner lab on its recent tests, all under Purity Focus. These are the figures we read in September 2026:

Lab	Published tests	Window	Pass rate on recent tests
BTLabs	2,411	Jul 2025 to Sep 2026	74% (last 200)
Krause Analytical	6,372	Apr 2025 to Sep 2026	64% (last 200)
Chromate	674	Dec 2024 to Jul 2026	40% (last 200)
Janoshik	195	Apr to Sep 2025	33% (last 150)

These rates mix different sellers and different periods, so they do not measure accuracy. Finnrick says it estimates each lab’s offset from batches that several labs tested. When we checked in September 2026, its lab pages did not publish those offsets.

Which standard, and which basis for mg

A pass under a ±35% fill window can be a fail under ±25%. The mg figure also depends on the lab’s basis: net peptide or total powder, and which reference standard. Two published single cases show the effect. In September 2025, graymarket compared results for one retatrutide batch: Janoshik and Chromate read about 30 mg, Krause 26 mg on two vials. In August 2026, Krysia’s Substack reported a tirzepatide batch labelled 60 mg: Chromate 71.63 and 70.99 mg, Accumark 73.72 and 75.44 mg, BTLabs 62.3 mg. Neither case proves a lab wrong; both show why the basis must be stated. Our guide to why labs disagree and the content per vial test explain the basis.

Who chose the vial

A vial a buyer picked at random says more about what buyers receive than a vial a seller chose. Finnrick marks seller-submitted tests as such, and its Launch with Finnrick page says sellers in that paid programme can review and approve results before they are published; once published, results cannot be removed. Certificates in the COA databases come mostly from sellers’ pages and lab portals, so they show the batches that clients, often sellers, chose to test and publish.

Whether you can open the original report

A number copied onto a website is only as good as the report behind it. Open the report on the lab’s own domain with its ID and key. Disclosed Labs found 92.9% of the COAs it checked against lab portals in May 2026 to be authentic. PeptideBenchmark tracks reused records and lookup links that no longer respond. Our guide on how to verify a peptide COA shows the steps.

Averages hide the spread

An average purity of 99.5% can hide a fail. Disclosed Labs shows the standard deviation of purity for each lab, not only the mean. It treats a very narrow spread as a warning sign: a lab loses 25 points if the standard deviation is under 0.15% across 30 or more COAs from fewer than 5 sellers, unless the range of test panels it reports explains it. Our guide to 99.9% purity results explains why.

Corrections matter as well. Finnrick keeps a public corrections log. In March 2026 it recorded a purity of 92.83% that turned out to be integration noise, and the lab reissued the report. Check whether a result you rely on was later corrected.

Peptide test results on Reddit, Discord and forums

Many results posted on Reddit, Discord and Telegram come from informal group tests. One organiser collects the money, picks a few vials at random from a group buy and sends them to a lab. The COA then goes to the group, and copies sometimes spread further. Our guides on how to run a peptide group test and how to get peptides tested cover the steps.

Group tests and forum threads share results as screenshots and PDFs. Ask for the report ID and key, open the report on the lab’s own site, and compare lot, cap colour and fill with your own vial. A result you cannot open yourself is a claim, not evidence. Our guide to reading a COA covers what to look for once it is open.

Where HK Analytical reports fit

HK Analytical reports are private by default, and there is no public list to browse. Whoever holds the private link, or the report ID and key, can open the report and check it at hkanalytical.com/verify. Every number comes with its raw data, the report is a signed record, and each report states how the client is related to HK Analytical. A correction makes a new version, and the old one is marked superseded. Each report also carries our photo of the vial as received, so you can compare it with your own. The example report shows the format. Quotes are on request (get a quote), and group tests are covered on the group testing page.