Mazdutide really is approved. Just not here, and not at the dose in the headline.

Why this note exists

Mazdutide is a once-weekly glucagon receptor and GLP-1 receptor dual agonist, developed by Eli Lilly with Innovent Biologics and sold in China under the brand name Xinermei.[5] In pepmg’s own price index, generated on August 5, 2026, 14 vendors carried a mazdutide listing — 16 listings in total.[15] That is a modest footprint today, and it is the reason to write now rather than later: the peer-reviewed result that will drive the marketing landed three days ago.

On August 4, 2026, JAMA published GLORY-2, the phase 3 trial of the 9 mg dose.[1] Most compounds sold as research chemicals never get a paper like this. Mazdutide now has four randomized phase 3 trials published across three of the highest-profile journals in science and medicine. The evidence problem here is not that the data are thin. It is that the data are specific, and the claims that travel are not.

A note on what kind of evidence this is: every efficacy figure below comes from randomized controlled trials in humans, not animal models. All were funded by Innovent Biologics, and — a point we return to — every registered phase 3 trial of mazdutide was run in China.[1][2][3][4][13]

Regulatory ledger

What was actually approved, and where

The Chinese weight-management indication covers adults with a BMI of 28 or higher, or 24 or higher with at least one weight-related comorbid condition, alongside diet control and increased physical activity.[7][5]

We queried FDA’s Drugs@FDA database on August 7, 2026 and found no approved application listing mazdutide as an active ingredient.[14] A search of ClinicalTrials.gov on the same date returned no registered phase 3 trial of mazdutide outside China; the registered work in the United States, the United Kingdom and Germany is phase 1 and phase 2.[13] An approval by China’s NMPA is a real regulatory decision by a real regulator. It is not an FDA approval, and it does not transfer.

GLORY-1: the doses that were approved

GLORY-1 was a phase 3, double-blind, placebo-controlled trial in China that randomly assigned 610 adults 1:1:1 to mazdutide 4 mg, mazdutide 6 mg, or placebo for 48 weeks. Participants had a BMI of at least 28, or 24 to under 28 with at least one weight-related coexisting condition. Mean baseline weight was 87.2 kg and mean BMI was 31.1.[2]

Mean percentage change in body weight from baseline at week 48. GLORY-1’s two primary endpoints were assessed at week 32 under a treatment-policy estimand — which assesses effects regardless of early discontinuation or the initiation of new antiobesity therapies — and gave −10.09%, −12.55% and +0.45%, with 73.9%, 82.0% and 10.5% of participants losing at least 5% of body weight (P<0.001 for all comparisons with placebo). The published abstract does not separately name an estimand for the week 48 figures.[2]

These are the doses a regulator has actually cleared, and they are the honest reference point for the name on a vial. The registry entry confirms the design — phase 3, 610 enrolled, all sites in China, completed — and lists no posted results, so the arm-level detail depends on the journal report.[12]

GLORY-2: one trial, three different headline numbers

GLORY-2 tested the unapproved 9 mg dose. It was a double-blind, placebo-controlled phase 3 trial at 27 hospitals in China, running from December 2023 to November 2025, randomizing participants 2:1 to 9 mg mazdutide or placebo for 60 weeks alongside a reduced-calorie diet and increased physical activity. Of 461 participants who received treatment, 64.0% were female, 16.1% had type 2 diabetes, mean age was 33.9 years, mean weight 94.0 kg and mean BMI 34.3.[1]

None of these numbers is fabricated, and they are not in conflict. They describe different populations and different analysis choices of the same trial. But “mazdutide gives 20% weight loss” quietly selects the largest of the three, from a subgroup, at a dose no regulator has approved — and drops the fact that the peer-reviewed whole-population figure is 16.65%. The registry entry for GLORY-2 lists 462 enrolled, quadruple masking, sites in China only, and no posted results.[11] Innovent announced in November 2025 that a supplementary application for the 9 mg dose had been accepted for review by China’s Center for Drug Evaluation; we found no announcement of a decision on that application at the time of writing.[10]

The diabetes trials, briefly

Two further phase 3 trials were published in Nature in April 2026, both in Chinese adults with type 2 diabetes, and both at the approved 4 mg and 6 mg doses.

DREAMS-1 randomized 320 participants 1:1:1 to mazdutide 4 mg, 6 mg, or placebo for 24 weeks, followed by a 24-week extension in which all participants received mazdutide. Glycated hemoglobin fell 1.57% and 2.15% against 0.14% on placebo (treatment differences −1.43% and −2.02%, both P<0.0001), with weight change of −5.61% and −7.81% against −1.26% (both P<0.0001).[3]

DREAMS-2 is the head-to-head. It randomized 731 participants on background oral antidiabetic drugs 1:1:1 to mazdutide 4 mg, mazdutide 6 mg, or dulaglutide 1.5 mg for 28 weeks. Both mazdutide doses were non-inferior and superior to dulaglutide on glycated hemoglobin, with least-squares mean treatment differences of −0.24% (P=0.0032) and −0.30% (P=0.0003) — small margins — and clearly greater weight reduction, −3.78% and −5.76% relative to dulaglutide (both P<0.0001). Gastrointestinal adverse events were more frequent with mazdutide.[4]

Dulaglutide 1.5 mg is a reasonable comparator in that setting, but it is not the strongest available agent. No published phase 3 trial has compared mazdutide with Semaglutide or Tirzepatide at their higher doses. Head-to-head trials against semaglutide are registered in China; we found no published results for them at the time of writing.[13]

The 9 mg result came with a gastrointestinal burden

In GLORY-2, the most common adverse events in the 9 mg group were vomiting in 53.1% of participants (against 1.3% on placebo), nausea in 46.9% (3.2%), and diarrhea in 39.4% (6.5%). The trial characterized most as mild to moderate, and adverse events leading to treatment discontinuation were reported in 2.9% of the mazdutide group against 0% on placebo.[1] The published conclusion states the benefit and the burden in the same sentence: clinically meaningful weight reduction, but gastrointestinal adverse reactions relative to placebo.[1]

In the separate GLORY-1 trial at the lower, approved doses, adverse events led to discontinuation in 1.5% of the 4 mg group and 0.5% of the 6 mg group, against 1.0% on placebo.[2] Those figures are not a head-to-head dose comparison with GLORY-2, but they make the 9 mg safety results important context for its larger weight-loss number. Both trials delivered mazdutide under supervision, with a defined titration schedule, a manufactured drug product and monitoring. None of those conditions transfers to a vial bought online.

What an independent review says about certainty

Everything above comes from the sponsor’s own trial programme. The most useful outside check is a network meta-analysis published in The BMJ on July 8, 2026, covering 262 randomized trials and 99,791 participants across 19 drugs, with certainty graded using GRADE.[6]

It found moderate-to-high certainty evidence of substantial weight loss at one year for tirzepatide (−14.9%), cagrilintide–semaglutide (−14.8%), oral semaglutide (−10.9%), orforglipron (−9.9%), subcutaneous semaglutide (−9.8%) and phentermine–topiramate (−8.1%). Mazdutide was not in that group. It was placed with ecnoglutide and Retatrutide among emerging agents that “may produce similar or greater reductions (13.1–14.6%)” — at very low to low certainty.[6]

That rating is not a claim the drug does not work; the phase 3 results are real and were published in serious journals. It is a statement about how much confidence the current evidence base supports, and the reasons are visible in this note: a single sponsor, a single country, no head-to-head against the strongest comparators, and no posted registry results for either GLORY trial.[11][12] Certainty and effect size are different axes, and marketing only ever quotes one of them.

Where the evidence does not reach

Four gaps sit between the trial record and a vial sold as a research chemical.

First, the population. Every phase 3 participant was recruited in China.[13] The GLORY-2 cohort had a mean age of 33.9 years and a mean BMI of 34.3; GLORY-1’s mean BMI was 31.1, and its entry criteria admitted participants from a BMI of 28 — below the BMI of 30 that GLORY-2 itself used to define obesity.[1][2] Whether these results generalize to other populations is a genuinely open question, not a rhetorical one — it is the kind of question a multi-region phase 3 programme exists to answer, and for mazdutide that programme has not reported.

Second, the dose. 4 mg and 6 mg are approved in China; 9 mg is not approved anywhere.[7][10] A vial’s milligram content is not a dose, and pepmg does not convert one into the other.

Third, the material. A trial result belongs to the specific manufactured drug substance that was tested — a known identity, content, impurity profile and sterility. A name match on a storefront establishes none of that, and pepmg does not certify what any vendor ships.

Fourth, the supervision. The trials titrated the dose, monitored participants and paired the drug with a reduced-calorie diet and increased physical activity in every arm, including placebo.[1][2] The reported effect is the effect of that whole package.

Questions people are asking