The five-seller comparison
“Most tested” here means the most distinct retatrutide laboratory reports in TitrateLab’s eligible international seller group. It is a ranking of the evidence we hold, frozen on September 6, 2026. We assessed reliability separately rather than awarding a reliability badge for having more reports.
The group contains 30 reviewed international seller identities and 1,730 eligible reports. The five leaders account for 677 reports. “International” uses TitrateLab’s reviewed supplier classification and excludes the U.S.-reseller group and unclassified identities. It does not verify a factory, a physical country of manufacture or which entity made a particular sample. The international supplier directory explains the broader cohort; the method below defines this retatrutide-only extraction.
| Rank | Seller identity | Reports, all time | Reports, last 90 days | Latest report |
|---|---|---|---|---|
| 1 | UTHER — Uther Peptide | 151 | 30 | August 5, 2026 |
| 2 | HK Peptides | 148 | 11 | August 13, 2026 |
| 3 | Lilipeptide | 145 | 13 | August 21, 2026 |
| 4 | QLP — Qing Li Peptide | 124 | 8 | August 13, 2026 |
| 5 | ZLZ — ZLZ Peptide | 109 | 11 | August 13, 2026 |
The 90-day window is June 9–September 6, 2026, inclusive. These are report dates, not dates of purchase or a claim about the sellers’ current inventory. Source: TitrateLab’s frozen report-level extraction. The live dossiers and retatrutide evidence hub may subsequently change and use different populations for their displayed totals.
Recent testing is quieter than the historical totals suggest
Across these five sellers, the archive contains 171 reports dated March 7–September 6, 2026, compared with 393 in September 7, 2025–March 6, 2026. All five have fewer reports in the latest six-month window, although each has an August-dated report.
| Seller | Previous six months | Latest six months | First → latest 92 days |
|---|---|---|---|
| UTHER | 72 | 53 | 22 → 31 |
| HK Peptides | 89 | 31 | 20 → 11 |
| Lilipeptide | 101 | 39 | 26 → 13 |
| Qing Li Peptide | 63 | 23 | 14 → 9 |
| ZLZ | 68 | 25 | 14 → 11 |
The equal 92-day periods are March 7–June 6 and June 7–September 6, 2026. UTHER’s report count increased between them; the other four decreased. The six-month calendar windows contain 184 and 181 days respectively. Source: a September 6 follow-up using the same report-selection method; all-time totals still matched the original snapshot.
What the testing says about reliability
High median purity does not erase an identity failure. The purity medians below include usable positive measurements and exclude missing results and records classified as identity concerns. Read the median together with the adverse-report column; it is not a summary of every submitted sample.
| Seller | Median reported purity | Results contributing to that median | Source-confirmed identity failures in the selected record |
|---|---|---|---|
| UTHER | 99.74% | 136 | 0 found |
| HK Peptides | 99.89% | 144 | 2 |
| Lilipeptide | 99.80% | 138 | 6 |
| Qing Li Peptide | 99.81% | 123 | 0 found |
| ZLZ | 99.85% | 106 | 1 |
“0 found” means no qualifying identity-failure signal appeared in this extraction. It does not mean every result explicitly confirmed identity. The available data do not provide a complete, consistently recorded denominator of identity assays, so we do not turn these counts into seller-wide failure percentages. All nine adverse findings were checked against their exact Finnrick public summaries, linked in the relevant seller sections below. Original laboratory PDFs were not reviewed for this article.
The five records also differ in sample-amount coverage and in what can be concluded from endotoxin entries. Those differences matter more than small differences between the purity medians.
1. UTHER: the largest report record, with unresolved testing questions
UTHER’s record leads this cohort with 151 reports, including 30 dated in the last 90 days. Its dated history in this extraction runs from August 20, 2024 to August 5, 2026. The positive-purity subset is tightly grouped: a 99.74% median across 136 results, ranging from 98.91% to 99.97%.
The amount evidence needs an additional check. 107 records have fields eligible for a measured-versus-label comparison, but two use a 90 mg reference against measurements near 26 mg. We could not corroborate that those reference amounts and measurements describe the same unit of product. We therefore leave those two label-to-sample matches unresolved and do not publish an underfill conclusion from them. The other comparison labels were not all independently rechecked either, so these fields do not establish a verified consistency rate.
One additional concern needs careful wording. Our archive stores a positive endotoxin signal for the June 6, 2026 sample. Finnrick’s public test history marks that exact test incomplete overall; the reviewed public material does not establish a quantitative endotoxin limit exceedance. We therefore report an unresolved signal, not a proven threshold failure.
2. HK Peptides: extensive coverage, with two confirmed adverse identity reports
HK Peptides ranks second with 148 reports, 11 in the last 90 days. Its 99.89% median across 144 positive-purity results is the highest of these five, but that figure excludes the most consequential adverse results.
Two distinct Finnrick test summaries dated January 29, 2026 explicitly report an identity failure: test q2ukhk2 and test ufgbcsa. We matched the exact test identifiers; another test on the same day is not interchangeable evidence. These summaries support a finding about the submitted samples, not a claim that every HK product has the same outcome.
Only 7 amount records have fields eligible for a label comparison, before independent label-to-sample verification. We excluded 134 supplier-batch-claim entries from that comparison. The apparently large testing record therefore does not provide equally large coverage of vial-label consistency.
3. Lilipeptide: the newest report in the five, and the most identity failures
Lilipeptide has 145 reports, including 13 in the last 90 days. Its latest eligible report, August 21, 2026, is the newest among these five. Positive-purity results have a 99.80% median across 138 measurements, but the range extends down to 89.26%, and 3 of those 138 measurements are below 97%.
The more important concern is six source-confirmed identity failures. Finnrick’s exact public summaries identify failures on October 17, 2025, January 7, January 15, January 29, May 11, and July 9, 2026.
Only 2 amount records have fields eligible for a label comparison, before independent label-to-sample verification. That is too small a subset to characterize general amount consistency. The other amount entries are mostly noncomparable batch claims, with identity concerns and missing information excluded as well.
4. Qing Li Peptide: broad report coverage, with unresolved endotoxin interpretation
Qing Li Peptide, shown as QLP in the dossier, ranks fourth with 124 reports, 8 in the last 90 days. Its positive-purity subset has a 99.81% median across 123 results. One result is below 97%, and the observed minimum is 95.63%.
No qualifying identity-failure signal appeared in this extraction. That is a narrower finding than saying that all 124 results explicitly established identity. The amount evidence is also limited: 122 batch-claim entries are excluded from label comparisons, leaving only 1 record with potentially comparable fields, before independent label-to-sample verification.
The archive’s positive endotoxin signal for the April 22, 2026 sample needs the same qualification as UTHER’s. Finnrick’s public test history marks that exact test incomplete overall. Without a numerical result, unit and applicable acceptance criterion, the reviewed public material does not establish an endotoxin limit failure.
5. ZLZ: mostly high positive-purity results, with one confirmed identity failure
ZLZ completes the five with 109 reports, including 11 in the last 90 days. Its positive-purity median is 99.85% across 106 results, with an observed range of 98.01–100%.
Its January 29, 2026 Finnrick test explicitly reports an identity failure. That adverse result is separate from the positive-purity distribution and must not disappear behind the median.
There are just 3 amount records with potentially comparable fields, before independent label-to-sample verification. A further 102 batch-claim entries are excluded from the label comparison. This is insufficient evidence for a broad claim about consistency against vial labels.
Why QSC and BFF are not in this top five
QSC was included in the comparison population. Qingdao Sigma Chemical ranks 15th with 52 eligible retatrutide reports, including 8 in the stated 90-day window. Its broader published record contains 63 retatrutide report groups; 11 lack a qualifying retained artifact or report reference under this ranking’s rules. The fifth-place cutoff is 109 eligible reports. QSC’s omission is therefore a report-count result, not a judgment that it is less reliable.
BFF exposes a classification gap in the comparison. Entries named Bfflink and Bfflist AMO do not have a reviewed international/U.S.-reseller classification in TitrateLab’s archive, so the international filter excludes them before ranking. That does not establish that they are domestic sellers or lack testing. A September 6 follow-up found 73 Bfflist AMO reports and 23 Bfflink reports passing the remaining evidence gates; neither separate count reaches the fifth-place cutoff. Finnrick also publishes separate testing histories for Bfflist AMO and Bfflink. We have not combined those identities or inferred a manufacturing location. Their absence limits the comparison’s coverage; it is not an adverse reliability finding.
What these records cannot establish
Report attribution is not a verified manufacturing chain. All five TitrateLab dossiers display a limited-attribution notice: the name’s role on the source reports has not been independently verified. The international cohort classification and the report association are useful organizing information, not independent proof of factory ownership, chain of custody or the origin of every sample. Janoshik’s terms, paraphrased from the Czech text, likewise distinguish customer-supplied sample information and measured results from product approval.
Testing dimensions are separate. A reported purity percentage does not supply an unreported sterility result, establish a numerical endotoxin limit, or prove the amount in another vial. Janoshik lists sterility, endotoxin and other analyses separately. An overall “incomplete” test classification also cannot be rewritten as an endotoxin-specific pass or fail.
The article does not measure fulfillment reliability. We did not analyze orders, shipments, refunds, inventory, or a representative buyer survey. Reports may reflect supplier-commissioned testing, repeat samples, targeted investigations or uneven collection. The amount of evidence available about a vendor is not its market share, and an absent report is not evidence that a test never happened.
A testing record is not regulatory approval. The FDA states that retatrutide is not a component of any FDA-approved drug. In its August 12, 2026 statement, Lilly says retatrutide remains investigational and no medicine containing it has approval from any regulator. This article compares documentary evidence; it makes no clinical-use recommendation.
Method and sources
Snapshot. TitrateLab extracted the data in a read-only transaction beginning September 6, 2026 at 00:42:12 UTC, using the report-eligibility and duplicate-handling methods serving the consumer site at that time. The source archive contained 16,049 published retatrutide rows before the international and evidence-quality filters. After selection, 1,759 rows belonged to 1,730 eligible reports across 30 international identities. These are different units, not inconsistent totals.
Eligibility. We selected the exact retatrutide compound group and the reviewed international supplier cohort. Records had to be published and attributed, with a retained artifact or a qualifying report reference. Flagged, suppressed, merged and unclassified vendor identities were excluded. A report associated with multiple vendors or unresolved attribution could not qualify. Exclusion is a limit of this comparison, not a finding about every excluded seller’s products.
Ranking. We ordered vendors by distinct eligible report count. The defined tie-breaks were retained-artifact report count, latest valid report date, then a stable canonical identity. No tie-break changed the five selected positions. One report may contain more than one result; copies do not become additional reports. The selected five contain 677 reports and 682 deduplicated analytical results. Of the reports, 616 have a retained artifact and 61 rely on a qualifying reference; this does not mean all 677 original laboratory documents were manually re-read for this article.
Measurements. Positive-purity summaries use finite values above zero and at or below 100%, excluding identity concerns. Distinct conflicting analytical outcomes are retained rather than silently collapsed. The below-97% band describes this analysis and is not a pharmaceutical specification. We excluded 510 Finnrick batch-claim amount entries from label comparisons, alongside missing, identity-concern or otherwise noncomparable entries. The remaining 120 amount records are candidates based on the available fields, not 120 independently verified label-to-sample matches; this article does not publish a seller-wide amount-consistency rate. Public sources use anonymous lab identifiers for some tests; we do not infer the identity or accreditation of those laboratories from the codes.
Source checks. We reviewed the exact public Finnrick summaries behind all 9 identity-failure signals and confirmed their stated identity outcomes. For the 2 positive endotoxin signals, the public histories show incomplete overall tests and do not establish quantitative exceedances. These checks reconcile source wording with the archive; they do not replace direct laboratory-document review or a complete audit of the corpus. Some TitrateLab report detail pages require membership; the external references in this article were publicly accessible when checked.
Process. This article used AI assistance for aggregate extraction, source reconciliation and drafting. The dated evidence snapshot, extraction method and source-check record are retained for editorial review.
For broader context, read the earlier, separately dated retatrutide purity and amount analysis, explore the live retatrutide COA hub, or consult TitrateLab’s methodology. To report a source, attribution or calculation error, use [email protected]. Material corrections will be recorded on this article.
