Gray Market Peptides 2026: What 9,997 Lab Tests Found in the Vials

About one in three peptide samples tested by the independent platform Finnrick fail basic quality checks, and that rate has barely moved in the 12 to 14 months of data behind it. The same dataset reads as high as 71.1% under stricter pass criteria, so the number depends on where the pass line sits. Most failures are dosing accuracy, not purity.

This page is built from primary artifacts checked on September 14, 2026: Finnrick’s live dataset, methodology, partner-lab and product pages; the Crossref DOI record for a preprint re-analysis of the same public data; the peer-reviewed JMIR purchase test of illegal online semaglutide; TitrateLab’s published certificate corpus; and the Guardian and New York Magazine reporting that first carried the number to a general audience. On gray-market peptide safety, the grey-market spelling used in British coverage describes the same trade measured by the same instruments. Every figure below carries its source and its date in the sentence where it appears. The September 15, 2026 section below adds WIRED’s purchase test of four vials bought from a Brooklyn bodega.

September 15, 2026: what the two new lab tests found

WIRED bought four injectable peptide products from a vendor called INDR Labs at the Mr. Green bodega on Bedford Avenue in Brooklyn, paid $376.20 for them, and sent the vials to Finnrick, the independent testing platform it describes as an Austin-based startup that brokers samples to outside laboratories. None of the four contained the substance named on its label: the BPC-157 vial, for example, had no BPC-157 in it. Heavy metals and endotoxins came back low, which WIRED reads as a sign the vials were probably not actively harmful, not as evidence that they were what the label promised.

Sources: WIRED, September 15, 2026 (rows 1-2, reporting Finnrick’s results); Peptide Critic’s lab report of August 13, 2026, as reported by WIRED (rows 3-4, reporting Analytical Formulations’ results). Both checked September 15, 2026.

The certificates of analysis were denied by the lab that supposedly issued them

WIRED reported that INDR Labs did not respond to its request for comment, but that shortly after it reached out, the company’s website began requiring login credentials, having already cut off public access to its certificates of analysis. The company claimed those certificates were provided by Vanguard Laboratory, and according to CBS News, which reported the denial on July 2, 2026, a spokesperson for Vanguard denied issuing the paperwork. Vanguard’s operations manager told CBS News: “We did not issue these [certificates], we do not have a client called INDR Labs”. Peptide Critic found the same gap in its own review: a QR code printed on the vials resolved to the vendor’s website rather than to a batch-specific certificate of analysis.

The corroborating test: a second vendor and a second lab

A month before WIRED published, Jeff Colhoun, the founder of the research-peptide vendor directory Peptide Critic, bought INDR Labs retatrutide and NAD+ from Smoker’s World, the smoke shop down the street from Mr. Green, and sent them to the Texas-based Analytical Formulations. That test is dated August 13, 2026: it is prior work WIRED reported, not a second commission, and the two results were not simultaneous. The laboratory could “not positively identify” any of the four vials — neither retatrutide sample matched its reference standard, neither NAD+ sample matched the NAD+ standard, and purity and quantity were reported indeterminate. Colhoun’s own statement of the limit is the sentence worth keeping: “These results do not prove that the vials were empty, and we are not claiming that they were”. The identity failure is the same class of risk as the counterfeit retatrutide case that tore an oesophagus.

What the result does and does not establish. Finnrick runs a label-claim analysis: the sample goes against a standard for the compound the sender says it is, and the answer is whether that compound was found. A negative therefore means the named compound was not detected — it does not mean the vial contained nothing, and it does not say what was in it. Marco Krause, the chief executive of the laboratory that performed the tests for Finnrick, put that limit on the record in WIRED’s reporting: “What we offer is simply a label claim analysis”, and, of what the vials contained, “Hard to say”. Two mechanisms fit a negative result and neither is established: the vial was mislabelled, or it once held the right peptide and was corrupted past recognition — peptides are sensitive to temperature and require refrigeration, so improper storage can degrade them. Finnrick’s chief executive, Raphaël Mazoyer, called the samples WIRED tested “just duds” while saying he cannot be certain what is going on. A label-claim failure differs from the market’s more common failure, a right molecule at the wrong dose, which our lab data on dose accuracy covers.

Where the gray-market vials come from

WIRED reports Mazoyer’s explanation for how a branded peptide vial reaches a bodega shelf: he points to the popularity of “turnkey services, which help pretty much anyone open a peptide business”, and to a system producing many labels from one source — “There are a whole bunch of labels popping up that have the same upstream vendor”. Finnrick has seen an uptick in these results over the past three months, he says, while noting his company has also been testing more samples as it has grown, and Colhoun told WIRED he has heard the same from vendors who test before selling. Buyers carry the most risk when sellers “skip meaningful batch testing” altogether. That mechanism — one upstream supplier, many labels, and a batch test a seller can elect to skip — is what why the research-chemical gray market is not going away describes. Demand-side, the same two channels appear in the only published count of BPC-157 users: of 1,039 clinician-documented users, sourcing was recorded for 205 of them, and 35% of those came from a gray-market vendor against 36% from a compounding pharmacy (BPC-157 real-world use data, nference analysis reported by Reuters, September 15, 2026). The analysis carries its own disclosure: its authors are employees of nference, a firm that runs research collaborations with biopharmaceutical companies whose products appear in the study, and it states that no external funding was received. The compounding half of that split is legal only under prescription, and the BPC-157 compound guide covers the April 2026 rule change and how to verify a pharmacy.

What 9,997 tested vials actually show

Finnrick’s public count is direct and current: “Customers sent us 9,997 peptide vials. Every result drives public safety rankings.” The homepage read on September 14, 2026 with its results marked “Last updated 3 Sept 2026”. Three compounds carry most of that volume.

Source: Finnrick homepage, checked September 14, 2026.

Finnrick is not the laboratory. Its methodology page states that it “sources samples from vendors and sends them to third-party labs”, and its partner-lab page credits each facility with the results it produced: Krause Analytical 6,160 tests (24 April 2025 to 26 August 2026), BTLabs 2,255, Chromate 674, Janoshik 195 and MZ Biolabs 179. The same page carries the platform’s own limit: “Finnrick does not assume that any single lab represents a ground truth.”

What counts as failing is explicit rather than impressionistic. In Purity focus, a sample passes at 99.5% purity or better and within plus or minus 35% of the labelled quantity. The Finnrick Standard tightens the dose window to plus or minus 25% and adds a batch-labelling score of at least 0.5. The members-only Accuracy focus uses 98% purity, a plus or minus 20% dose window and a 40 EU per sample endotoxin ceiling. Underfill and overfill both fail when they exceed the allowed range, a failed identity result fails the sample outright, and published evidence carries a 730.5-day half-life, so an old test counts about half as much as a new one.

Where the “one in three” number comes from (and what it does not say)

Finnrick published the headline itself on February 26, 2026, in a company post signed by chief executive Raphaël Mazoyer: “We’ve tested over 7,000 peptide samples. Roughly one in three fail basic expectations.” The Guardian carried it on April 6, 2026, in a report by Sarah Marsh and Nicola Davis: “about a third of the thousands of products it analysed failed basic quality checks, and that proportion had stayed broadly unchanged over the 12 to 14 months it had been gathering data.” New York Magazine’s January 2026 feature quoted BioTools co-founder Rina Dukor at a softer figure: “Nearly 30 percent of the peptides tested by Finnrick are either mislabeled, under- or overdosed, or contaminated with toxins or foreign bacteria.”

Those are not three measurements of the same quantity. Finnrick’s number is the share of samples missing at least one stated pass line. New York Magazine’s folds identity, dose and contamination into one looser ratio. And the Guardian’s restatement arrives with a threshold the platform does not use: the “98% threshold” for purity is the newspaper’s wording, not Finnrick’s, whose live Purity focus passes at 99.5% and reserves 98% for the members-only Accuracy mode. Quote the number without its source attached and it will not mean what you think.

Same dataset, three numbers. The variable is the pass line, not the measurement. Ask only whether a vial missed any stated expectation and roughly a third fail. Tighten the dose window, add the batch-labelling requirement, and score every dimension separately so a sample can fail on any one of them, and the same corpus returns 41.6% to 71.1%. That spread is not a disagreement about what the laboratories measured; it is three questions asked of one spreadsheet, which is why neither figure should be quoted without the other in the same breath.

The harsher numbers come from a second pass over that data. A preprint by Mendias and Awan, posted April 24, 2026 (DOI 10.20944/preprints202604.1748.v1) and not peer-reviewed, screened 6,441 peptide samples across fourteen compounds and reported: “Between the two models, 41.6% to 71.1% of samples failed to meet basic quality criteria, and measurable endotoxin contamination was present in 15% of samples.” Its limitations section treats the sample as non-representative of the market and the failure rate as a lower bound.

Who benefits from which number is worth knowing. Partnership for Safe Medicines, an advocacy group with a PhRMA-adjacent funding lineage, revived the figure in July 2026: “Cases like these are especially troubling given reports from a Texas lab that a third of the peptide samples they test fail identity, purity, or quantity checks”. PeptidesDirect, which sells peptides, argues the other way from the same corpus — median purity 99.80% (interquartile range 99.50 to 99.90), median measured abundance 101.80% of label (IQR 95.00 to 109.00), identity failures in 2.4% of samples. Both readings can be arithmetically correct, one optimising for reassurance and the other for alarm, and neither changes what the underlying measurement is.

What a failing vial actually fails on

For anyone weighing the risks of buying research peptides online, this is where the risk physically sits. The three categories Finnrick uses are identity, purity and quantity, and they fail at very different rates. The one buyers ask about least is the one that fails most.

Identity: the vial is a different molecule

In the re-analysis corpus, 2.4% of samples were not the compound on the label, with much higher rates for specific compounds: TB-500 at 10.0%, CJC-1295 at 9.1%, tesamorelin at 3.6% and BPC-157 at 3.5%. Identity failures are not always subtle. New York Magazine reported vials labelled semaglutide that contained retatrutide, at least one that contained no active peptide at all, and unlabelled glucose filler, which is a specific hazard for anyone with diabetes because filler that never appears on the label can be read as a dose of something else.

The pattern is older than the peptide boom. A 2017 study in JAMA of 44 products sold online as selective androgen receptor modulators found only 23 (52%) contained a SARM, 17 (39%) another unapproved drug and 4 (9%) no active compound; a 2015 study in Drug Testing and Analysis found melanotan II vials from three shops measuring 4.32 to 8.84 mg against a 10 mg claim. Both are older work and neither represents today’s peptide market, but the failure mode is identical: unregulated e-commerce mislabels, and the buyer cannot detect it by looking. The September 15, 2026 update above carries the newest identity failures: four bodega vials that did not hold the compound on the label.

Purity: usually high, and the weakest reassurance on the page

On the number most buyers check first, the market looks better than the headlines suggest. Median purity in the re-analysis corpus was 99.80%. TitrateLab, a peptide-data desk that reviewed 62,079 published certificates, reports that “Nearly three of every four certificates come back with ≥99.5% purity”, with the monthly sub-95% fail rate falling “from about 6.6% in early 2025 to under 1% by mid-2026”. TitrateLab is a market participant and we attribute those numbers to it for that reason.

Purity is a ratio, and a ratio can look excellent in a vial that is wrong in every other way. The peer-reviewed JMIR 2024 purchase test of semaglutide from illegal online pharmacies found measured purity “between 7.7% and 14.37%”, against labels claiming 99%. Purity also says nothing about what the other 90-odd percent of the powder is, or whether the powder is sterile.

Quantity: where the corpus actually fails

Dosing accuracy is the dominant deficit in every public dataset. The re-analysis puts it starkly: “For tirzepatide the pattern was even sharper: only 13.8 percent passed both criteria, while 58.8 percent met purity and failed abundance” — the right molecule, mostly pure, and the wrong amount. Median abundance across the corpus was 101.80% of label, meaning half of all vials carried more drug than advertised, and an interquartile range reaching 109.00% means a quarter carried at least 9% more.

The Australian data we covered in August found the same shape in counterfeit retatrutide: of 18 samples tested this year by Leeder Analytical: “Roughly 30 percent contained higher concentrations than stated, 30 percent contained less, and one vial contained no retatrutide at all.” A vial that is 30% over-strength is not the safer version of a vial that is 30% under-strength. It is the same missing quantity check, with the sign reversed.

Endotoxin, and what a standard panel never looks at

The 15% endotoxin figure is real but thin. It rests on 243 samples out of 6,487 screened, under four percent of the corpus, because endotoxin is an add-on order rather than a default test. Where it has been measured, the results are not reassuring: Finnrick fails a sample above 40 EU per sample when testing is ordered, and the JMIR semaglutide purchase test detected endotoxin in every vial it received, at levels between 2.1645 and 8.9511 EU per mg.

Sterility, particulate matter, residual solvents and stability are not measured by a standard certificate of analysis at all. Those are the columns a 99% purity figure cannot speak to, and they are the ones that decide whether an injectable is safe rather than merely potent.

Semaglutide Contamination Testing 2026: What a Buyer Can Actually Order

For a lyophilised GLP-1 vial, contamination is not one question. It is five. Is the powder the molecule on the label (identity)? Is the peptide intact and free of truncated or aggregated variants (purity)? Does the vial contain the mass the label claims (quantity)? Is there bacterial endotoxin, which drives fever and immune reactions? And is the product sterile and free of particulates? A peptide contamination test answers as many of those as it was ordered to answer, and no more.

Source: Finnrick’s semaglutide testing page, checked September 14, 2026. A sample can also be submitted anonymously, without vendor details, for $120.

Two things follow. The most common result is also the cheapest one: most vials that reach a laboratory pass identity, purity and quantity, and the corpus-wide failure rate is driven by a minority. And the add-ons mark exactly where the blank spaces are — endotoxin and heavy metals are known only when someone pays to test for them, so a clean cheap panel is three answers out of five, not a safety certificate.

The peer-reviewed semaglutide evidence is small and stark. Ashraf and colleagues purchased semaglutide from illegal online pharmacies, had three injection vials delivered, and found measured purity of 7.7% to 14.37% against labels claiming 99%, content over label by 28.56% to 38.69%, and endotoxin in all of them. Three vials cannot be generalised to a market, but this is the only peer-reviewed purchase test of the drug we could find, and it points the opposite way from the certificate feeds.

Where mislabelling enters is specific rather than general. Most failures are dose, not identity; the identity failures that do happen are the ones that matter most. A vial labelled semaglutide that contains retatrutide — an unapproved triple agonist owned by Eli Lilly, in Phase 3 trials and approved nowhere — delivers a different drug, at a different strength, from the one the buyer believes they have. Our tracker covers retatrutide’s approval status, and the physical tells are in how to spot fake pens and vials.

How to Verify a Peptide Certificate of Analysis

A certificate of analysis is a laboratory report about one sample. It is not a property of a vendor, a batch or a product line, and the gap between those two things is where buyer confidence goes wrong. On peptide purity testing cost the numbers are published and small; on peptide certificate of analysis content the standards are checkable.

What a COA has to contain

Four elements make a certificate checkable. An identity method — mass spectrometry or an equivalent — rather than a purity percentage alone. A purity figure with the method named, which in practice means HPLC. A quantity in milligrams, not only a percentage. And a batch identifier that also appears on the vial in your hand. That last element is the one vendors drop most often, and Finnrick’s labelling score is built on precisely that distinction: points for a batch identifier printed on the container, and none for one that exists only on a website.

Six questions, which is how to verify peptide purity claims without a laboratory of your own

1. Who paid for the test? A certificate commissioned by the seller is a seller’s document, however clean it looks.
2. Does the batch identifier on the certificate match the vial you are holding? A batch ID that exists only on a web page identifies nothing.
3. Is the quantity stated in milligrams, or only as a purity percentage? A purity ratio cannot detect an underfilled vial.
4. Is the testing laboratory named, with a report number you can look up?
5. Does the figure appear in an independent public record — a published dataset — rather than only on the vendor’s own site?
6. Does it cover sterility and endotoxin, or only identity, purity and quantity?

Six clean answers make a certificate worth reading. Fewer makes it marketing. The September 15, 2026 tests show what a certificate that was never issued looks like in practice: the lab named on the vendor’s certificates denied issuing them.

What testing costs

On Finnrick’s semaglutide panel, identity, purity and quantity testing is free; endotoxin adds $110; a heavy-metals panel adds $120; and an anonymous submission without vendor details costs $120. Those are the only prices cited on this page and they are the platform’s own published figures, not a market average.

Which peptide testing labs are publishing results

Finnrick’s partner-lab page names five facilities with published counts, led by Krause Analytical at 6,160 tests (24 April 2025 to 26 August 2026), then BTLabs at 2,255, Chromate at 674, Janoshik at 195 and MZ Biolabs at 179. Those laboratories do not agree with one another on the same samples: TitrateLab’s review of its corpus found Finnrick averaging +2.1% against label where Janoshik reads +7.4% and Freedom Diagnostics +10.1%, which TitrateLab calls “a methodology disagreement, not noise”. That is why a few percentage points of difference between two certificates is not automatically a red flag, while a label-to-vial gap of 30% is.

How other countries verify peptide vials

Verification is not only a buyer-side problem. Regulators are building the same check from the other end: a serial number on the box, and a record of who imported and sold it. Thailand’s September 2026 rules show the shape.

Thailand: prescription-only from September 15, 2026

Thailand’s Public Health Ministry issued Notification No. 58, placing injectable GLP-1 medicines — liraglutide and semaglutide, the weight-loss pens — in the specially controlled category. From Tuesday, September 15, 2026, pharmacy sales require a doctor’s prescription and dispensing must be by a pharmacist. Thailand’s FDA framed the change as access-preserving: patients who need treatment are not the target, and it aims at unsupervised weight-loss use and online sales. This is Thai law governing dispensing in Thailand, not a US rule. The safety context: the STEP 1 semaglutide trial’s 2022 extension found participants regained about two-thirds of their lost weight within a year of stopping, in that trial population.

Track & Trace and the serial-number check

Thailand’s FDA has asked licensed importers and sellers of GLP-1 medicines to report sales data into its Track & Trace system, so a medicine can be followed along the chain — importation, storage, distribution, dispensing to the patient — and traced back when irregularities appear. Phase one is a request for cooperation, not a binding obligation, and the detailed conditions are still being drafted. Patients can enter the box serial number into the FDA’s pilot serialisation menu; the FDA advises avoiding a product when no authorisation record appears, or the details do not match the packaging. See how to verify a legitimate peptide source in our FAQ.

The enforcement backdrop: unregistered vials seized near Bangkok

On September 11, 2026, a joint police and FDA operation at a rented house in Bang Yai district, Nonthaburi, seized unregistered injectable peptide products sold online for weight loss — peptide vials, mixing liquid, syringes and packaging — worth more than 52 million baht. The Thai FDA described the haul as unregistered products rather than a confirmed seizure of counterfeit branded GLP-1 pens; our report on the same week’s Australian counterfeit-vial case documents it. Thai law attaches prison terms and fines to selling specially controlled medicines without a prescription, to trading drugs without a licence and to selling unregistered medicines; advertising them unlawfully can draw a fine of up to 100,000 baht.

What you can and cannot verify from the packaging

For a US buyer, the packaging signals that matter are a serial or lot number tied to a traceable record, a batch identifier matching the vial, and a manufacturer you can look up. Where none exists, there is no verification path at all.

What a serial number check can confirm

• That the box appears in the Track & Trace record as an authorised product through licensed importation and distribution.
• A mismatch, which is the signal: the FDA advises avoiding the product when the details do not match the packaging.

What it cannot confirm

• What is inside the vial. Identity, concentration, endotoxin and sterility are not what serialisation measures.
• Anything about a vial decanted, repackaged or refilled: serialisation verifies the box, not the vial.
• Anything outside the issuing jurisdiction: the Thai pilot says nothing about a box bought in the US.

What it means when there is no verification path at all

• No regulator-side check exists, so the only evidence left is laboratory testing of the sample you send, which characterises the sample rather than the batch. Our 10-point pharmacy verification checklist covers what a seller’s documents should show.
• Absence is not proof of a fake, and presence is not proof of safety: the box can be vouched for, the vial stays unverified. See how to spot counterfeit retatrutide pens and vials. For GLP-1 specifically, twenty state attorneys general asked the White House in September 2026 to treat Chinese-sourced GLP-1 ingredients as a national-security problem.

Sources: Nation Thailand — Thailand makes GLP-1 injections prescription-only from September 15 (September 14, 2026); Nation Thailand — Thai FDA reports over 52m-baht peptide haul on September 11 (September 12, 2026). Both read September 14, 2026.

Comparison: what each public dataset measures

Each row below is a different instrument answering a different question. No competitor page on the measured search results carries this comparison.

Read together, the four agree on the shape and disagree on the size. Purity is mostly high and improving. Quantity is where failures concentrate. Identity failures are rare and, when they happen, the most dangerous. And the measurements nobody orders — sterility, particulates, residual solvents, stability — are the ones no dataset can speak to. The comparison does not answer whether Finnrick is legit, which is a reputational question no single dataset can settle, but it shows precisely what each source can and cannot support.

What nobody knows yet

Six limits bound everything above, and they belong in the argument rather than in a footnote.

• Sampling is voluntary. Samples reach these laboratories because a buyer, a vendor or the platform chose to send them. Vendors confident in their product may over-submit; vendors with a problem have no reason to submit at all.
• The re-analysis is a preprint. It has not been peer-reviewed, and it is a second pass over data somebody else collected.
• Endotoxin coverage is under 4% of the corpus — 243 of 6,487 samples screened. A 15% finding on that base is a signal, not a rate.
• Sterility, particulates, residual solvents and stability are effectively unmeasured across every public dataset on this page.
• Laboratories disagree with each other by 2 to 10 percentage points on dose, so a result close to a threshold says something about the assay as well as the vial.
• The direction is well evidenced; the magnitude is not. The most honest sentence available comes from a seller: PeptidesDirect states that the real failure rate for modern research peptides is not scientifically established. We would put it slightly differently — the failures are real and reproducible in kind, and their exact size is an open question.

If you already have gray-market vials

Three things, and none of them is a dosing instruction.

• Do not inject a vial whose contents have not been verified. Every failure mode above — the wrong molecule, the wrong amount, bacterial contamination — is invisible in the vial.
• If you want to know what is in it, have it tested, or dispose of it safely. Testing characterises the sample you sent, not the batch you bought, and it is not a permission slip for use.
• Severe or worsening abdominal pain, or signs of infection at an injection site, warrants urgent medical care. New York Magazine’s January 2026 feature reported a man who used retatrutide for several months, developed increasingly severe stomach pain, and was diagnosed with acute necrotizing pancreatitis; the same piece reported two women hospitalised in Las Vegas in July 2025 after injecting peptides at a health fair.

We are not your clinician. The harm this market causes is documented in hospital records, including patients hospitalised after counterfeit injections, a counterfeit retatrutide case that tore an oesophagus and an online-sourced retatrutide case report. On enforcement we track FDA warning letters to peptide vendors and an endotoxin recall of compounded peptides; for the same failure mode in a smaller dataset, see our lab data on dose accuracy. The ER survey of gray-market use covers who uses these products and the ECRI and ISMP finding the purity range US safety groups documented. Our sourcing standards govern every number here; about PeptidesBeat explains who we are.

Frequently asked questions

Are gray-market peptides safe?

About one in three peptide samples tested by the independent platform Finnrick fail basic quality checks, and that rate has barely moved in the 12 to 14 months of data behind it. The same dataset reads as high as 71.1% under stricter pass criteria, so the number depends on where the pass line sits. Most failures are dosing accuracy, not purity. The split in the data is the part that matters: purity is usually high while quantity frequently is not, so a certificate reporting 99% pure answers only one of the five questions worth asking about a vial.

What percentage of gray-market peptides fail testing?

Roughly one in three. Finnrick’s own published count, dated 26 February 2026, is that it has tested over 7,000 peptide samples and that roughly one in three fail basic expectations; the Guardian reported in April 2026 that about a third of the products analysed had failed, and that the proportion had stayed broadly unchanged over the 12 to 14 months of data behind it. Higher figures exist: a preprint re-analysis of the same public data, not peer-reviewed, scores 41.6% to 71.1% of samples as failing depending on where the pass line is drawn, with measurable endotoxin in 15% of those screened. That spread is one dataset read at different thresholds, not three separate findings.

What is the difference between purity and content in a peptide COA?

Purity is a ratio: how much of the powder is intact peptide rather than related impurities. Content, or abundance, is a mass: how many milligrams are actually in the vial. A half-full vial of very clean powder can still report 99% purity. That is why the same re-analysis corpus shows a median purity of 99.80% and, at the same time, only 13.8% of tirzepatide samples passing both criteria while 58.8% met purity and failed the dose check.

How do I test semaglutide for contamination, and what does it cost?

Finnrick lists semaglutide testing at no charge, with identity, purity and quantity measured by HPLC and the sample running at Krause Analytical in Austin, Texas or BTLabs in Florida. Endotoxin testing is an add-on at $110, a heavy-metals panel covering lead, arsenic, cadmium and mercury is an add-on at $120, and an anonymous submission without vendor details costs $120. Sterility, particulates, residual solvents and stability are not part of a standard panel and must be requested separately, which is why most public results do not cover them.

Is Finnrick legit, and what is a Finnrick rating worth?

It publishes its method, names its partner laboratories with their test counts, and states plainly that it does not assume any single lab represents a ground truth, which is more disclosure than most rating services offer. Three limits bound it. Its samples are voluntary submissions rather than a random sample of the market, so its failure rate is not a market average. Its own pages disagree about where samples come from. And laboratories disagree with each other by 2 to 10 percentage points on dose, so a rating near a threshold partly reflects which lab ran the assay. A passing rating is evidence about one sample, not proof that a product is safe.

What do certificates of analysis not cover?

Sterility, particulates, residual solvents and stability are generally absent unless a buyer commissions them, and even endotoxin and heavy metals are add-ons rather than defaults: endotoxin data existed for 243 of 6,487 screened samples in the re-analysis corpus, under four percent. A certificate reporting identity, purity and quantity has told you three things about one vial and nothing about the batch it came from, how it was shipped, or what happens to the powder over time.

How do I check a semaglutide serial number?

Thailand's FDA operates a pilot serialisation service: a box serial number is entered into the Track & Trace menu on the FDA website. A record means an authorised product reported by a licensed importer or seller. The FDA advises avoiding the product when no authorisation record appears, or the details do not match the packaging.

That covers the box, not the vial: identity, concentration, sterility and endotoxin are not what serialisation measures. Report a null or mismatched result on hotline 1556 or the Oryor Smart Application.

Do Thailand's GLP-1 prescription rules apply to US buyers?

No. Notification No. 58 is Thai law and governs dispensing inside Thailand from September 15, 2026, so it does not reach a buyer in the United States. It matters to US readers as a model: it shows the verification signals a regulator can attach to a GLP-1 box, and what their absence means.

The Thai instrument covers only boxes that passed through licensed Thai channels, so it gives a US reader a model to look for, not a check to run. Nothing in the United States changes as a result.

Do bodega peptides contain what the label says?

Not the four vials WIRED bought from INDR Labs at a Brooklyn bodega for $376.20: none of the four contained the compound named on its label, and the BPC-157 vial contained no BPC-157. Finnrick runs a label-claim analysis, so the result shows the stated compound was absent — it does not show what the vial did contain. Low heavy metals and endotoxins is a statement about those two measures, not a statement that the vials were what they claimed to be.

What is a certificate of analysis and can it be faked?

A certificate of analysis is a laboratory report about one sample, and there are three ways it can mislead: the report never existed, it was issued for a different batch, or it was never issued by the laboratory it names. The September 15, 2026 evidence documents the third: the company named on the paperwork claimed its certificates came from Vanguard Laboratory, CBS News reported that Vanguard denied issuing them, and the vendor stopped making certificates public. A certificate is only as strong as the laboratory’s willingness to confirm the report number on it, and the test underneath it is still a label-claim check on one sample — it says whether that sample matched the compound it was submitted as, not what the rest of the batch holds. Check the batch identifier on the certificate against the vial in your hand; how to spot fake pens and vials covers the physical tells.

If the vial is a dud, was it mislabeled or degraded?

Both fit the data, and a label-claim test cannot separate them: it says whether the stated compound was present, not why it was missing. A vial filled with something else and a vial whose peptide broke down in heat or in transit both come back as a negative result, so the finding is evidence about the sample rather than a verdict on the cause. Peptides are temperature-sensitive and require refrigeration, which is why improper storage remains a live explanation alongside mislabelling — and neither WIRED’s reporting nor the two laboratories established which applies to these vials.

Sources

1. New York Magazine — Life on Peptides Feels Amazing (2026-01-27) — source of the nearly-30-percent figure and the semaglutide-for-retatrutide mislabelling reporting.
2. The Guardian — labs testing thousands of unregulated substances amid peptide craze (2026-04-06) — the about-a-third attribution and the three failure categories.
3. Finnrick — live dataset and vendor rankings — 9,997 published tests, last updated 3 Sept 2026; per-compound depth.
4. Finnrick — testing methodology — pass lines, labelling score, endotoxin ceiling, 730.5-day evidence half-life, sample sourcing.
5. Finnrick — What We Believe In (2026-02-26) — the platform’s own roughly-one-in-three statement, signed by its chief executive.
6. Finnrick — partner testing laboratories — per-lab published test counts and the ground-truth caveat.
7. Finnrick — free semaglutide testing — panel composition, add-on prices and the laboratories a semaglutide sample may run at.
8. Mendias C.L., Awan T.M. — preprint re-analysis of the same public data (MDPI Preprints, posted 2026-04-24; not peer-reviewed) — 6,441 samples, 41.6% to 71.1% failure, 15% endotoxin.
9. PeptidesDirect — One in Three Peptides Fails Testing? (2026-07-24) — median purity and abundance, identity rate, endotoxin coverage, and the vendor-side counter-reading.
10. TitrateLab — peptide grey-market certificate corpus (reviewed 2026-07-31) — 62,079 certificates, purity trend, lab-to-lab disagreement. A market participant; attributed in-line.
11. Ashraf et al., JMIR 2024 (DOI 10.2196/65440) — peer-reviewed purchase test of semaglutide from illegal online pharmacies: three vials, 7.7% to 14.37% purity, endotoxin in all.
12. Van Wagoner et al., JAMA 2017 (DOI 10.1001/jama.2017.17069) — historical comparator: 44 products sold online as SARMs.
13. Breindahl et al., Drug Testing and Analysis 2015 (DOI 10.1002/dta.1655) — historical comparator: melanotan II vials 4.32 to 8.84 mg against a 10 mg claim.
14. Partnership for Safe Medicines — Five reasons (2026-07) — the advocacy restatement of the one-in-three figure.
15. PeptidesBeat — editorial policy — our sourcing and preprint-labelling rules.
16. PeptidesBeat — Lab Tests Show Black-Market Peptides Miss the Labeled Dose (2026-08-28) — the Australian dose-accuracy dataset cited above.
17. WIRED — Those Viral Bodega Peptides Aren’t Actually Peptides (2026-09-15) — the four INDR Labs vials bought at a Brooklyn bodega, the Finnrick label-claim results, the Vanguard Laboratory denial, and the Peptide Critic corroboration.

Corrections. This page is compiled from primary sources, each dated where it appears. Where a figure is live — Finnrick’s published test count in particular — the date it was read is printed with it. If a figure changes after publication we annotate the change and its date here rather than editing it silently. Last verified September 15, 2026.

PeptidesBeat is an independent editorial publication covering peptide policy, research, and industry developments. We do not sell peptides, recommend dosing, or provide medical advice. All content is informational. Peptides referenced may be subject to FDA restrictions; consult a licensed healthcare professional for any therapeutic question.