Mazdutide vs Survodutide vs Tirzepatide: The Glucagon Receptor Everyone Said Would Raise Blood Sugar, the China-Only Approval, and Why the Press-Release Numbers Never Match the Journals

Mazdutide vs Survodutide vs Tirzepatide: The Glucagon Receptor Everyone Said Would Raise Blood Sugar, the China-Only Approval, and Why the Press-Release Numbers Never Match the Journals

For a decade the objection to putting a glucagon receptor agonist in a weight-loss drug was obvious to anyone who had taken first-year physiology: glucagon raises blood sugar. The counter-argument was that glucagon also raises energy expenditure and burns liver fat, and that enough GLP-1 activity in the same molecule would cover the glucose. In 2026 the argument is being answered by data rather than by physiology. Mazdutide lowered HbA1c by 2.15 percentage points against 0.14 for placebo in treatment-naive diabetics; survodutide lowered it by 0.8 to 0.9 points against 0.2 over 76 weeks by the journal analysis. On average, HbA1c fell rather than rose in every trial reported, and the reason it was added, the liver, showed up exactly where it should: liver-fat reductions of 63 to 73 percent with mazdutide and a 30 percent or greater reduction in 84 percent of survodutide patients. What the field has not settled is whether the extra receptor buys more weight loss than tirzepatide's GIP arm, and here the comparison is harder than any headline suggests. Mazdutide's pivotal trials enrolled Chinese adults with a mean BMI of 31 to 34; tirzepatide's and survodutide's enrolled multinational cohorts at BMI 38; retatrutide's at BMI 40. The companies lead their press releases with the efficacy estimand, the journals lead with the treatment-regimen estimand, and the gap between the two is three to four percentage points for every drug with high discontinuation. Survodutide lost 16.6 percent in the press release and 13.0 percent in the New England Journal; those are the same trial. This reference lays out the molecules, every pivotal number with its estimand attached, the tolerability picture that has split the two dual agonists, and the regulatory map as of October 2026, in which one of them is a prescription drug in China and both appear on the FDA's list of drugs it has warned companies for illegally selling under "research purposes" labels in the United States.

Michael Phelps

Founder & Peptide Research Specialist, PrymaLab

Research reference · Last updated October 4, 2026 · ~38 min read

Four molecules, three receptors: what each one is

The four peptides in this article are variations on one engineering idea: take a gut or pancreatic hormone, swap residues so it survives in the blood, hang a fatty acid off it so it rides on albumin for a week, and tune which of three receptors it hits. The receptors are the GLP-1 receptor, which suppresses appetite, slows gastric emptying and drives glucose-dependent insulin release; the GIP receptor, the other incretin, whose contribution to weight loss is still debated; and the glucagon receptor, which raises hepatic glucose output but also raises energy expenditure and drives fat oxidation in the liver.^[24][46] How a molecule balances those three defines what it is.

Mazdutide: an oxyntomodulin with a tail

Oxyntomodulin is a 37-amino-acid product of the proglucagon gene, made in the gut after meals; it contains the full 29-residue glucagon sequence plus an eight-residue C-terminal extension, and in its native form it activates both the glucagon and GLP-1 receptors weakly and is destroyed within minutes.^[24] Mazdutide's developers describe it as "a long-acting synthetic peptide analogue of mammalian oxyntomodulin (OXM), with a fatty-acyl moiety to extend the half-life," which "potently bound to human and mouse GLP-1 receptors and glucagon receptors in vitro."^[16] Public structural summaries give a C20 fatty-acid side chain, molecular formula C₂₁₀H₃₂₂N₄₆O₆₇ and a mass of about 4,563 Da.^[59] I was not able to confirm the residue count, the acylation position or a peer-reviewed receptor potency ratio from a primary source; Innovent's papers describe the agonism only as "balanced."^[13][16] In mice it reduced body weight in both glucagon-receptor-knockout and GLP-1-receptor-knockout animals, meaning each arm works on its own, and it increased energy expenditure.^[16] In a phase 1b study in Chinese patients with type 2 diabetes, absorption was slow with a median Tmax of about 72 hours and the terminal half-life ranged from 147 to 673 hours, six to 28 days, across participants; four of 24 developed anti-drug antibodies, none neutralizing.^[16] The molecule was discovered by Eli Lilly as LY3305677, licensed for China to Innovent Biologics as IBI362, and Lilly retains rights everywhere else.^[22]

Survodutide: a glucagon taught to behave like GLP-1

Survodutide came from the opposite direction. Zimmermann and colleagues at Boehringer Ingelheim describe BI 456906 as "a 29-amino-acid peptide optimized to balance the degree of agonism for the human GCGR and GLP-1R by introducing various modifications to the human glucagon sequence."^[24] Position 2, where DPP-4 would cleave it, carries the non-coded amino acid 1-aminocyclobutane-1-carboxylic acid; the C-terminus is amidated; and "a glycine-serine linker in position 24, carrying a C18 di-acid," provides the albumin binding that extends the half-life.^[24] In CHO cells the cAMP EC50 was 0.33 nM at the GLP-1 receptor and 0.52 nM at the glucagon receptor, roughly 5- and 26-fold less potent than native GLP-1 (60 pM) and glucagon (20 pM) respectively but near-equipotent at the two targets.^[24] In diet-induced obese mice it produced greater weight loss than a maximal dose of semaglutide, "driven by increased energy expenditure and reduced food intake," with glucagon-receptor engagement confirmed by hepatic gene expression, circulating amino acids and FGF-21.^[24] It was discovered at Zealand Pharma and is developed by Boehringer.^[34]

Tirzepatide and retatrutide: the GIP-first designs

Tirzepatide is "a fatty acid modified peptide with dual GIP and GLP-1 receptor agonist activity," 39 residues on a GIP backbone with a C20 diacid at lysine 20 and a half-life of about five days.^[43][44] Retatrutide extends the same scaffold to a third receptor: "LY3437943 shows balanced GCGR and GLP-1R activity but more GIPR activity," with Lilly's own figures giving GIPR potency 8.9 times that of GIP, GLP-1R potency 2.5 times lower than GLP-1, and GCGR potency 2.9 times lower than glucagon, and a half-life of about six days.^[46][47] The retatrutide versus tirzepatide reference covers that pair in detail; here they are the yardsticks.

How survodutide was chosen from 19 dual agonists

Survodutide was one of 19 dual glucagon receptor and GLP-1 receptor agonists put through a biomarker-led selection programme, described by Thomas and colleagues in 2024 (Thomas et al. 2024). The team measured cAMP potency in CHO-K1 cells stably expressing the human glucagon and GLP-1 receptors, and checked activity at receptors expressed naturally, using mouse MIN6 insulinoma cells for the GLP-1 receptor and rat primary hepatocytes for the glucagon receptor (Thomas et al. 2024). In mice, improvement in oral glucose tolerance was the read-out for GLP-1 receptor engagement, while rises in plasma fibroblast growth factor-21 (FGF21) and in liver nicotinamide N-methyltransferase (NNMT) mRNA were the read-outs for glucagon receptor engagement (Thomas et al. 2024).

After acute exposure in lean mice, these target engagement markers correlated significantly with the in vitro potencies across all 19 compounds (Thomas et al. 2024). Three candidates, survodutide, BI 456908 and BI 456897, went on to fuller pharmacological profiling, with body weight studied in diet-induced obese mice and glucose lowering in diabetic db/db mice, and survodutide was qualified as the clinical development candidate (Thomas et al. 2024). The liver NNMT marker also appears in the discovery paper by Zimmermann and colleagues as evidence that BI 456906 engages the glucagon receptor in vivo (Zimmermann et al. 2022). The selection work therefore rests on mouse and rat systems, and the abstracts do not give the potency ratio between the two receptors that set survodutide apart from the other 18 compounds.

Why add glucagon, and what happened to blood sugar

The rationale is stated plainly in the survodutide discovery paper. Oxyntomodulin "reduces bodyweight by increasing energy expenditure and reducing energy intake in humans," and "at pharmacological doses, glucagon increases energy expenditure." But "glucagon is the key counter-regulatory hormone" to insulin, so the design goal is to pair glucagon-receptor agonism with enough GLP-1-receptor agonism to cancel the glucose rise.^[24] Lilly's retatrutide paper frames the same trade: glucagon-receptor-mediated energy expenditure is added on top of incretin-driven intake reduction.^[46] In mice all three glucagon-containing molecules raised energy expenditure, and survodutide outperformed maximally effective semaglutide.^[16][24][46]

Two things follow in the clinic. The first is the liver. Glucagon-receptor agonism drives hepatic lipid oxidation, and liver fat is where the dual agonists look most different from everything else. In GLORY-1, participants with at least five percent liver fat at baseline lost 63.26 percent of it on 4 mg and 73.18 percent on 6 mg, against an 8.20 percent gain on placebo, at 48 weeks.^[2] In GLORY-2, participants without diabetes and with at least 10 percent baseline liver fat lost 71.9 percent on 9 mg against a 5.1 percent gain.^[5] Survodutide's SYNCHRONIZE-MASLD trial reached a 30 percent or greater liver-fat reduction in 84.2 percent of participants against 24.3 percent on placebo by the efficacy estimand, and its phase 2 MASH study produced histological improvement in 62 percent at 4.8 mg against 14 percent on placebo.^[27][36]

The second is glucose, and the fear did not materialise. In treatment-naive Chinese adults with type 2 diabetes, mazdutide 6 mg lowered HbA1c by 2.15 percentage points against 0.14 for placebo at 24 weeks.^[9] In SYNCHRONIZE-2, survodutide lowered it by 0.9 points on 3.6 mg and 0.8 on 6.0 mg against 0.2 on placebo over 76 weeks by the journal analysis, and by up to 1.21 against 0.03 by the efficacy estimand.^[32][33] The developers attribute this to GLP-1-receptor-driven insulin secretion and weight loss outweighing glucagon-driven hepatic glucose output.^[24] The mazdutide 9 and 10 mg phase 1b study recorded no hypoglycaemia at all.^[14] What the glucagon arm did cost, at least early, was heart rate: in mazdutide's phase 2 the 24-week change was +5.82, +5.38 and +8.75 beats per minute at 3, 4.5 and 6 mg against +4.81 on placebo, and the authors noted that earlier dual agonists cotadutide, SAR425899 and JNJ-64565111 had produced rises of 5 to 15 bpm.^[13] By 48 weeks in GLORY-1 the increase was +2.6 bpm in both active arms.^[2] Retatrutide's heart-rate rise peaked at 24 weeks and declined thereafter.^[45]

One caveat the whole field shares: none of the pivotal trials measured human energy expenditure by calorimetry. The thermogenic contribution of glucagon in people is inferred from mice and from the shape of the weight curves, not measured.^[24][46]

Mazdutide: GLORY, DREAMS and the first approval in the class

Mazdutide is the only GLP-1/glucagon dual agonist that is a licensed medicine anywhere, and its entire pivotal record is Chinese. That matters for both the numbers and their interpretation.

Phase 2

The 24-week phase 2 in 248 Chinese adults with overweight or obesity compared three dose levels with placebo and reported weight changes of −6.7, −10.4 and −11.3 percent against a 1.0 percent gain on placebo, with early discontinuation of 10.2 versus 8.1 percent and no drug-related serious adverse events.^[13] A second part, in 80 participants at 9 mg, reported a placebo-adjusted 18.6 percent loss at 48 weeks, with liver fat down 73.3 percent at 24 weeks and ALT down 45.5 percent against placebo in those with fatty liver; that 48-week result circulated through company announcement and conference, and I could not find a peer-reviewed 48-week paper.^[15] A separate 20-week phase 2 in type 2 diabetes lowered HbA1c by 1.41 to 1.67 points against 1.35 for dulaglutide 1.5 mg.^[17]

GLORY-1, the approval trial

GLORY-1 was published in the New England Journal of Medicine in 2025. Six hundred and ten Chinese adults aged 18 to 75 with a BMI of at least 28, or 24 to 28 with a weight-related condition, and without diabetes, were randomised to mazdutide 4 mg, mazdutide 6 mg or placebo for 48 weeks. Baseline weight was 87.2 kg and BMI 31.1.^[1][3] At the week-32 primary endpoint, by the treatment-policy estimand, weight fell 10.09 percent on 4 mg and 12.55 percent on 6 mg against a 0.45 percent gain on placebo; 73.9, 82.0 and 10.5 percent lost at least five percent.^[1] At week 48 the efficacy estimand gave −12.05 and −14.84 percent against −0.47, with half the 6 mg arm losing 15 percent or more, waist down 9.48 and 10.96 cm, and blood pressure, lipids, uric acid and transaminases all improved.^[2] The most frequent adverse events were nausea, diarrhoea and vomiting, mostly mild or moderate.^[2] China's NMPA approved mazdutide for chronic weight management on 27 June 2025, for adults with a BMI of 28 or more, or 24 or more with a comorbidity, the Chinese thresholds.^[18][19]

GLORY-2, the 9 mg trial

GLORY-2 moved to the higher dose and a heavier population. Four hundred and sixty-two Chinese adults (461 treated) with a BMI of at least 30, 16.1 percent of them with type 2 diabetes, baseline weight 94.0 kg and BMI 34.3, were randomised two to one to 9 mg or placebo for 60 weeks across 27 hospitals.^[4][6] The JAMA paper, published in August 2026, reports −16.65 percent (95 percent CI −18.19 to −15.12) against −1.50 percent, a 15.15-point difference, in participants who received at least one dose.^[4] Innovent's efficacy-estimand figures were −18.55 against −3.02 percent overall and −20.08 in participants without diabetes, with 44.0 percent losing 20 percent or more, liver fat down 71.9 percent in the fatty-liver subgroup, and discontinuation for adverse events in 2.9 percent against none on placebo, a figure the JAMA abstract also reports.^[4][5] The JAMA authors' conclusion was measured: clinically meaningful weight reduction, "but participants receiving the drug experienced gastrointestinal adverse reactions."^[4] A supplementary application for 9 mg was accepted by the NMPA on 24 November 2025; I found no approval announcement in sources through early October 2026.^[21]

DREAMS: the diabetes programme

Two phase 3 diabetes trials appeared back to back in Nature on 17 December 2025. DREAMS-1 randomised 320 treatment-naive adults, baseline HbA1c 8.24 percent and BMI 28.2, to 4 mg, 6 mg or placebo for 24 weeks: HbA1c fell 1.57 and 2.15 points against 0.14, and the 6 mg arm lost 7.81 percent of body weight.^[9] DREAMS-2 randomised 731 adults on oral agents to mazdutide 4 mg, 6 mg or dulaglutide 1.5 mg for 28 weeks and showed HbA1c non-inferior and then superior to dulaglutide, with differences of −0.24 and −0.30 points, and weight differences of −3.78 and −5.76 percent, at the cost of more gastrointestinal events.^[10] Those two trials earned the second NMPA approval, for glycaemic control in type 2 diabetes, on 19 September 2025.^[20] DREAMS-3, an open-label head-to-head in 349 adults with early diabetes and obesity, reported in October 2025 that 48.0 percent on mazdutide 6 mg against 21.0 percent on semaglutide 1 mg reached both HbA1c under 7 percent and 10 percent weight loss at 32 weeks, with weight down 10.29 against 6.00 percent.^[11] Two caveats attach: the comparator was semaglutide at its diabetes strength, not its higher obesity strength, and the trial was open-label and is published only as a design paper.^[12] The trial that would answer the obesity question properly, GLORY-3 against semaglutide 2.4 mg in 479 adults with obesity and fatty liver, is active and not recruiting with primary completion estimated for 31 December 2026. No GLORY-3 results exist, and any circulating online should be treated as unverified.^[7][8]

Survodutide: SYNCHRONIZE, the liver, and a tolerability problem

Survodutide arrived at phase 3 with the stronger liver story and left it with the weaker weight story, and the reason is tolerability.

Phase 2: a signal and a warning

The phase 2 obesity trial, published in Lancet Diabetes and Endocrinology in 2024, randomised 387 adults with a BMI of 27 or more across 43 centres in 12 countries to 0.6, 2.4, 3.6 or 4.8 mg survodutide or placebo for 46 weeks. Weight fell 6.2, 12.5, 13.2 and 14.9 percent against 2.8 percent.^[25] Only 233 of 386 treated participants, 60.4 percent, completed the 46 weeks; those who reached and stayed on 4.8 mg lost 18.7 percent in an actual-treatment analysis; and discontinuation for adverse events ran at about 25 percent against 4 percent on placebo, mostly during escalation.^[25][26] The MASH phase 2 published in the New England Journal the same year was the one that drew attention: in 293 adults with biopsy-confirmed MASH and F1 to F3 fibrosis, 48 weeks of 2.4, 4.8 or 6.0 mg produced histological improvement without worsening fibrosis in 47, 62 and 43 percent against 14 percent on placebo by the planned-treatment analysis. The widely quoted "83 percent versus 18.2 percent" figure is the on-treatment analysis from the topline release.^[27][28] That result brought FDA Breakthrough Therapy designation for non-cirrhotic MASH with F2 to F3 fibrosis in 2024, on top of Fast Track from 2021.^[38]

SYNCHRONIZE-1 and -2

SYNCHRONIZE-1, published in the New England Journal in June 2026, randomised 726 adults without diabetes (725 analysed), BMI 30 or more or 27 with a complication, baseline weight 108.8 kg and BMI 37.9, to survodutide 3.6 mg, survodutide 6.0 mg or placebo for 76 weeks. By the treatment-regimen estimand, which "incorporates the effects of any early discontinuation, protocol-prohibited obesity medications, and a prolonged dose-escalation period," weight fell 12.2 percent on 3.6 mg and 13.0 percent on 6.0 mg against 5.4 percent on placebo; 72.6, 71.9 and 46.3 percent lost at least five percent.^[29][35] Boehringer's efficacy-estimand figures were up to 16.6 percent against 3.2, with 85.1 percent five-percent responders.^[30] Two features of that result stand out: the placebo response of 5.4 percent is unusually large, and the two active doses produced nearly identical results.^[29] Body-composition data presented at ADA 2026 showed visceral fat down 34 percent against 12 and liver fat down 63.1 percent against 25, with lean mass making up no more than 10.8 percent of tissue lost.^[31] Discontinuation for adverse events, as reported by the trade press from the ADA presentation, was about 24 percent on 3.6 mg and 25 percent on 6.0 mg against 5.4 percent on placebo, with vomiting in 45 percent on the high dose.^[31][34]

SYNCHRONIZE-2, in 752 adults with type 2 diabetes and a BMI of 27 or more, baseline HbA1c 7.4 percent, was published on 1 October 2026 alongside its EASD presentation. Weight fell 8.2 and 9.8 percent against 3.9 by the treatment-regimen estimand, up to 13.1 percent by the efficacy estimand, and HbA1c fell 0.9 and 0.8 points against 0.2 by the journal analysis, up to 1.21 against 0.03 by the efficacy estimand.^[32][33] Trade reports gave vomiting in 39 percent and adverse-event discontinuation of 26 percent; one analyst called the profile "less competitive vs other agents" and Zealand's shares fell 11 percent on the day.^[34] Those discontinuation and vomiting figures reach me through press coverage of the presentations and should be verified against the journal supplements.

The liver programme

SYNCHRONIZE-MASLD, in Nature Medicine in June 2026, randomised 216 adults with obesity and at-risk MASLD two to one to 6.0 mg or placebo for 48 weeks: 84.2 percent against 24.3 achieved a 30 percent or greater liver-fat reduction by the efficacy estimand, 68.5 against 28.6 by treatment-regimen, with liver fat normalising in about 61 percent against 5.7.^[36][39] The phase 3 MASH programme, LIVERAGE in about 1,800 adults with F2 to F3 fibrosis and LIVERAGE-Cirrhosis in about 1,590 with compensated cirrhosis, runs to roughly 2031.^[38][39] A Chinese phase 3, SYNCHRONIZE-CN, is testing 3.6 and 4.8 mg against placebo in 306 adults at the Chinese BMI thresholds over 76 weeks, with results pending.^[40] The cardiovascular outcomes trial, SYNCHRONIZE-CVOT, randomised 5,508 participants across 34 countries and is listed as completed on 30 June 2026 with no results public as this is written.^[37] Survodutide is approved nowhere, holds FDA Fast Track and Breakthrough designations, EMA PRIME and NMPA Breakthrough status, and Boehringer has not announced a filing date.^[30][34][38]

The benchmarks: tirzepatide and retatrutide in 2026

Any comparison needs the approved standard and the molecule everyone is waiting for. Tirzepatide's SURMOUNT-1 randomised 2,539 adults without diabetes, baseline weight 104.8 kg and BMI 38.0, to 5, 10 or 15 mg or placebo for 72 weeks; by the treatment-regimen estimand weight fell 15.0, 19.5 and 20.9 percent against 3.1.^[41] SURMOUNT-5, open-label, put maximum-tolerated tirzepatide against maximum-tolerated semaglutide in 751 adults for 72 weeks and found 20.2 against 13.7 percent.^[42] Tirzepatide is approved by the FDA as Mounjaro for diabetes (May 2022) and Zepbound for weight management (November 2023) and sleep apnoea (December 2024), by the EMA, and in China for both indications in 2024.^[19][44] The tirzepatide reference has the full record.

Retatrutide's phase 2 produced 24.2 percent loss at 48 weeks on 12 mg.^[45] The phase 3 results landed in 2026. TRIUMPH-1, in the New England Journal on 29 September 2026, randomised 2,339 adults without diabetes to 4, 9 or 12 mg or placebo for 80 weeks: by treatment-regimen, −17.6, −23.7 and −25.0 percent against −3.9, with co-primary endpoints in knee osteoarthritis pain and sleep apnoea also met.^[48] Lilly's efficacy-estimand figures were up to −28.3 percent, with a 104-week extension in 532 participants with a BMI of 35 or more reaching −30.3 percent; adverse-event discontinuation was 4.1, 6.9 and 11.3 percent against 4.9; and a new adverse event, dysesthesia, appeared in 12.5 percent against 0.9.^[49] TRIUMPH-2, in the Lancet the same week, found −18.8 percent at 12 mg in 1,152 adults with type 2 diabetes, and TRIUMPH-3 reported −22.6 percent in adults with established cardiovascular disease by the efficacy estimand in topline results (TRIUMPH-3 registry enrolment about 1,946).^[50][51][52] Lilly has said it will submit in the US in the first quarter of 2027.^[51] The retatrutide versus tirzepatide reference covers the molecule in depth.

Mazdutide vs tirzepatide vs survodutide vs retatrutide: the honest comparison

Here is the table everyone wants, with the columns that make it honest.

Read the placebo-adjusted row and the population row together. Survodutide's placebo-adjusted loss of 7.6 points is less than half of tirzepatide's, but its placebo arm lost 5.4 percent where tirzepatide's lost 3.1, and its treatment-regimen estimand counts the quarter of participants who stopped. Mazdutide's 15.2 points look close to tirzepatide's 17.8, but its participants started 10 kg lighter at a BMI four units lower, were followed 12 weeks less, and were all enrolled in China; percentage weight loss does not transfer across those baselines, and a 2026 systematic review of mazdutide notes the generalisability problem directly.^[23] Retatrutide's number is the largest in the class and comes with the heaviest population and the longest follow-up. A 2026 network meta-analysis of 58 trials in BMJ Medicine made the indirect comparison formally and found "low certainty evidence" for higher discontinuation with retatrutide, "whereas mazdutide showed better tolerability."^[53] That is as far as anyone can go without a head-to-head, and no head-to-head among these four has been run.

Mazdutide vs tirzepatide

No trial compares them. Mazdutide's best journal figure is 16.65 percent at 60 weeks in a BMI-34 Chinese cohort; tirzepatide's is 20.9 percent at 72 weeks in a BMI-38 multinational cohort.^[4][41] Mazdutide's reported adverse-event discontinuation is lower, 2.9 against 6.2 percent at tirzepatide's 15 mg dose.^[5][41] Mazdutide has a liver-fat dataset tirzepatide's primary trials did not report; tirzepatide has outcomes data, a cardiovascular programme and approvals on three continents. The only thing that would settle "which is better" is a trial neither company has announced.

Mazdutide vs retatrutide

Both are Lilly molecules, and Lilly's decision to develop retatrutide globally while licensing mazdutide to China is itself a data point. Retatrutide 12 mg: −25.0 percent treatment-regimen, −28.3 efficacy, at 80 weeks, with 11.3 percent stopping for adverse events and dysesthesia in one in eight.^[48][49] Mazdutide 9 mg: −16.65 and −18.55 at 60 weeks, 2.9 percent stopping.^[4][5] The triple agonist loses more weight and more patients.

Survodutide vs retatrutide, and survodutide vs mazdutide

Survodutide 6.0 mg against retatrutide 12 mg is 13.0 against 25.0 percent by the journal estimands, with discontinuation of about 25 against 11.3 percent, in populations of similar baseline BMI; on every axis the triple agonist leads.^{[29][31][48][49]} Survodutide against mazdutide is the within-class comparison: same receptors, opposite problems. Mazdutide lost more weight at a lower discontinuation rate in a lighter population; survodutide has the biopsy-proven MASH data, the dedicated MASLD trial and a completed cardiovascular outcomes trial that mazdutide does not.^{[4][5][27][29][36][37]} If the question is weight, the indirect data lean toward mazdutide; if it is liver histology, they favour survodutide; on adverse-event discontinuation, mazdutide by a wide margin, although vomiting was frequent with both.

Why the press-release number never matches the journal

Every drug in this article has two headline numbers, and the difference is not spin, it is statistics. The treatment-regimen or treatment-policy estimand asks what happened to everyone randomised, including people who stopped the drug, took prohibited medications or never reached the target dose. The efficacy estimand, sometimes called hypothetical or on-treatment, asks what would have happened had everyone stayed on the drug as planned. Journals lead with the first; press releases lead with the second. When almost nobody drops out the two agree; when a quarter of participants stop, they diverge sharply. SYNCHRONIZE-1: 16.6 percent in the press release, 13.0 in the journal.^[29][30] SYNCHRONIZE-2: 13.1 against 9.8.^[32][33] TRIUMPH-1: 28.3 against 25.0.^[48][49] GLORY-2: 18.55 against 16.65.^[4][5] A reader comparing mazdutide's 18.55 to tirzepatide's 20.9 is comparing an efficacy estimand to a treatment-regimen one and should stop. The same applies to placebo: GLORY-1's placebo arm gained 0.45 percent, SYNCHRONIZE-1's lost 5.4, so placebo-adjusted differences move with the trial as much as with the drug.^[1][29] And milligram figures do not translate: the four drugs are different molecules with different potencies and half-lives, so the arm labels say nothing about relative strength.

Trial design and tolerability

The pivotal trials raised each investigational drug in steps before participants reached their assigned dose, and that escalation period is where much of the tolerability record sits.^{[3][4][9][10][11][25][27][29][32][41][48][49]} Slower escalation or a lower starting dose was the main lever programmes used for tolerability (retatrutide's phase 2 found a lower start reduced GI events), and survodutide's discontinuation did not improve despite a "prolonged" schedule: 24.6 percent stopped in phase 2 and about 25 percent in phase 3.^[25][31] The headline results were reached only by participants who tolerated the climb; the efficacy estimand describes them, the treatment-regimen estimand describes everyone. PrymaLab's Mazdutide 10mg and Survodutide 10mg vials are research formats; the trials above describe other people's work with pharmaceutical products, not a use for them. The reconstitution guide covers laboratory handling of lyophilized peptides generally.

Safety: GI events, heart rate, and the discontinuation gap

Gastrointestinal events dominate for all four drugs and cluster during escalation.^{[1][25][29][45]} Retatrutide's TRIUMPH-1 rates at 12 mg were nausea 42.4 percent against 14.8, vomiting 25.3 against 4.8, diarrhoea 32.0 against 13.5, constipation 26.1 against 10.9, plus dysesthesia in 12.5 against 0.9.^[49] Vomiting was frequent with both dual agonists: 53.1 percent against 1.3 with mazdutide 9 mg in GLORY-2, alongside nausea in 46.9 against 3.2 and diarrhoea in 39.4 against 6.5, and 45 percent with survodutide 6.0 mg in SYNCHRONIZE-1 and 39 percent in SYNCHRONIZE-2 by press report.^[4][31][34] Where they differ is how many people stopped. Survodutide's adverse-event discontinuation, about 25 percent at 6.0 mg and 24 percent at 3.6 mg in SYNCHRONIZE-1 and 26 percent in SYNCHRONIZE-2, is essentially unchanged from phase 2.^[25][31][34] Mazdutide's was 2.9 percent at 9 mg in GLORY-2, 1.5 and 0.5 percent at 4 and 6 mg in GLORY-1 against 1.0 on placebo, and none drug-related in phase 2.^[1][4][13] Tirzepatide's SURMOUNT-1 discontinuation was 6.2 percent at 15 mg (4.3 to 7.1 across doses) against 2.6.^[41]

Heart rate is the glucagon-specific question and the data are incomplete. Mazdutide: +5.4 to +8.8 bpm at 24 weeks in phase 2, +2.6 bpm at 48 weeks in GLORY-1.^[2][13] Retatrutide: dose-dependent increase peaking at 24 weeks and declining.^[45] Survodutide: systolic blood pressure fell 6.2 mmHg placebo-corrected in phase 2, but heart-rate data by dose are not in the summaries I could access.^[26] Transaminases fell rather than rose with mazdutide, consistent with the liver-fat effect, and no hepatotoxicity signal appears in the cited trials.^[2][15] Hypoglycaemia was absent in the mazdutide 9 and 10 mg phase 1b; rates in the diabetes trials, where background insulin or sulfonylureas matter, were not extracted here.^[14] Both companies report no new safety signals beyond the GLP-1 class, whose approved labels carry warnings for rodent thyroid C-cell tumours, pancreatitis and gallbladder disease; whether the Chinese mazdutide label carries identical warnings is something I could not confirm from the insert.^[5][20][30]

Regulatory status: China, the FDA's October 2026 statement, and WADA

Mazdutide is approved by China's NMPA for chronic weight management in adults with a BMI of 28 or more, or 24 or more with a weight-related comorbidity, since 27 June 2025, and for glycaemic control in type 2 diabetes since 19 September 2025.^[18][20] The 9 mg supplementary application was accepted on 24 November 2025 and had not been approved in any source I found.^[21] It is not approved by the FDA or EMA; Lilly states that it "retains the rights to develop mazdutide outside of China and continues to advance its clinical development globally," listing obesity and alcohol use disorder as study areas, but no US phase 3 has been announced.^[22] A Chinese trade name circulates online but I could not confirm it from an Innovent or NMPA document.

Survodutide is approved nowhere. It holds FDA Fast Track (2021), FDA Breakthrough Therapy for non-cirrhotic MASH with F2 to F3 fibrosis (2024), EMA PRIME (November 2023), NMPA Breakthrough (June 2024) and Taiwan FDA Breakthrough (September 2024).^[38][58] Boehringer has not dated a submission.^[34]

Tirzepatide is approved in the US, EU, China and elsewhere; retatrutide is approved nowhere, with a US submission planned for the first quarter of 2027.^[19][44][51]

The US grey market. On 1 October 2026 the FDA updated its statement "FDA's Concerns with Unapproved GLP-1 Drugs Used for Weight Loss." It warns against products "falsely labeled 'for research purposes' or 'not for human consumption'" that are sold with human dosing instructions; reports 990 adverse-event reports for compounded semaglutide and more than 730 for compounded tirzepatide as of 31 May 2026; states that retatrutide and cagrilintide cannot legally be compounded; and states that the "FDA has warned companies that have illegally sold unapproved drugs containing semaglutide, tirzepatide, retatrutide, survodutide or mazdutide that are falsely labeled 'for research purposes' or 'not for human consumption,'" adding that "these products have been sold directly to consumers for human use with dosing instructions."^[54] On 24 August 2026 the agency sent warning letters to five peptide websites, stating that "despite statements on your product labelling marketing your products 'for research use only'... your products are intended to be drugs for human use."^[55] A Public Citizen review published in June 2026 counted more than 100 companies selling retatrutide online, found that 71 of 179 vendors tested by a third-party laboratory rated poor, bad or fraudulent, and tallied 26 FDA adverse-event reports for retatrutide in the first five months of 2026, including 11 hospitalisations.^[56] The regulatory line is intended use, not the label. PrymaLab sells mazdutide and survodutide as research materials with certificates of analysis, this article gives no dosing information, and it describes clinical trials of pharmaceutical products, not a use for research-grade peptide; the supplier verification guide explains what the certificate should show.

WADA. GLP-1 receptor agonists, including semaglutide and tirzepatide, are not prohibited on the 2026 list or the 2027 list effective 1 January 2027; semaglutide has been on the Monitoring Program since 2024 and tirzepatide has been added to it, and WADA's expert group found insufficient evidence to conclude with confidence that these substances meet the criteria for inclusion, while it considers a possible weight-management category.^[57] Mazdutide, survodutide and retatrutide are not named. Because survodutide and retatrutide are approved nowhere, they may fall under S0 (non-approved substances); mazdutide's Chinese approval takes it outside S0's wording.

Research-grade mazdutide and survodutide: what to verify

Three points for laboratories handling these peptides.

Identity is harder than purity. Mazdutide is about 4,563 Da and survodutide is a lipidated 29-mer with a non-coded residue; both are close in mass and chromatographic behaviour to other acylated proglucagon-family peptides. An HPLC trace shows how much of one thing is present, not which thing it is. A certificate should carry mass-spectrometric identity against the expected monoisotopic mass, and the purity testing guide explains why sequence confirmation matters more for lipidated peptides than for short ones.

Lipidated peptides behave differently in solution. The C18 and C20 fatty-acid chains that bind albumin in vivo drive self-association in vitro. Reconstituted solutions should be made in the buffer the experiment requires, kept cold, not repeatedly frozen and thawed, and used within the window the certificate specifies; the storage guide and reconstitution guide cover the general principles.

Know what the data describe. Every efficacy number in this article comes from pharmaceutical-grade product in a monitored trial. The FDA's October 2026 statement is a reminder that research labelling does not change the legal analysis if the material is intended for people. Research-grade vials are for laboratory work on receptor pharmacology, formulation, analytical method development and the like.

What this article does not settle

• Which is better. No head-to-head trial among mazdutide, survodutide, tirzepatide and retatrutide exists; everything above is indirect.^[53]
• Mazdutide outside China. All phase 3 data are from Chinese adults at BMI 28 to 34; generalisability to other ethnic groups and heavier populations is untested.^[23]
• GLORY-3. The only trial against semaglutide 2.4 mg in obesity reads out after December 2026.^[7]
• Survodutide's future. Whether Boehringer files for obesity with 13 percent treatment-regimen loss and 25 percent discontinuation, and what SYNCHRONIZE-CVOT showed.^[29][34][37]
• Energy expenditure in humans. Never measured in a pivotal trial; the glucagon thermogenesis story is still a mouse story.^[24][46]
• Durability. No withdrawal or maintenance trial for either dual agonist.
• Long-term heart rate and cardiovascular effects of chronic glucagon-receptor agonism.^[13][45]
• Lean mass. The "no more than 10.8 percent" figure awaits peer review and comparison with tirzepatide and retatrutide DXA sub-studies.^[31]
• Molecular detail. Mazdutide's residue count, acylation site and receptor potency ratio were not confirmable from primary sources.

Mazdutide, IBI362, survodutide, BI 456906 and "dual agonist": what the names refer to

Mazdutide is the international nonproprietary name; IBI362 is Innovent's code and LY3305677 Lilly's, and early papers call it simply an oxyntomodulin analogue.^[16][22] Survodutide is the INN for BI 456906, Boehringer Ingelheim's code for the Zealand-discovered molecule.^[24][34] Tirzepatide is LY3298176, sold as Mounjaro for diabetes and Zepbound for obesity and sleep apnoea.^[43][44] Retatrutide is LY3437943, "triple G" in the lay press. The term GLP-1/glucagon dual agonist, also written GLP-1R/GCGR co-agonist or glucagon-based dual agonist, covers mazdutide and survodutide along with discontinued or earlier molecules such as cotadutide, efinopegdutide and pemvidutide; dual agonist on its own is ambiguous, since tirzepatide is also a dual agonist of a different pair. The trial acronyms: GLORY and DREAMS for mazdutide; SYNCHRONIZE and LIVERAGE for survodutide; SURMOUNT and SURPASS for tirzepatide; TRIUMPH for retatrutide.

The approved comparators, and why their labels keep moving

Three of the four molecules in this comparison are investigational and one is approved only in China, so most readers arrive having used or read about a different drug entirely. The approved GLP-1 medications that frame the whole conversation are worth naming precisely, because the brand names carry different indications and doses even when the molecule is the same.

Semaglutide is sold as Ozempic for type 2 diabetes, as Wegovy for chronic weight management at a higher dose, and as Rybelsus in an oral tablet for type 2 diabetes. One molecule, three products, three dose ranges, and the Wegovy indication is the one people mean when they talk about weight loss. Semaglutide is a pure glucagon-like peptide-1 receptor agonist: one receptor, no glucagon arm, no glucose-dependent insulinotropic polypeptide arm.

Tirzepatide is the GIP and GLP-1 dual agonist and its label has moved twice recently in ways that matter for any comparison. Zepbound now carries two indications: chronic weight management, and moderate to severe obstructive sleep apnea in adults with obesity, the latter approved in December 2024. Mounjaro covers type 2 diabetes including paediatric patients aged 10 and over, and from late August 2026 also carries an indication for reducing major adverse cardiovascular events in adults with type 2 diabetes at high risk, on the SURPASS-CVOT result. Any table that lists tirzepatide as a weight-loss drug alone is now out of date in three separate ways.

That matters for reading the dual-agonist data. Mazdutide is FDA-approved nowhere; it is approved by China's NMPA, for chronic weight management from June 2025 at a body mass index threshold of 28, or 24 with a weight-related comorbidity, and for type 2 diabetes from September 2025. Survodutide and retatrutide are FDA-approved for nothing at all. A reader comparing a Chinese-approved drug and two investigational ones against a molecule whose own label gained a cardiovascular outcome indication in 2026 is comparing things at four different stages of evidence.

Blood sugar, insulin sensitivity and the glucagon problem

Adding a glucagon receptor agonist to a GLP-1 agonist creates a specific tension that does not arise with tirzepatide, and it is the most interesting pharmacology in this comparison. Glucagon raises hepatic glucose output. That is its job. So a drug that deliberately activates the glucagon receptor in people with impaired glucose handling is pushing in two directions at once, and the question every dual-agonist programme had to answer was whether the GLP-1 arm wins.

It does, in the trials run so far. Mazdutide lowered HbA1c in the DREAMS programme and beat dulaglutide on both HbA1c and weight in DREAMS-2, and survodutide lowered HbA1c by up to 1.21 percentage points in SYNCHRONIZE-2 from a baseline of 7.4 per cent, with 29.5 per cent of participants reaching a level below the pre-diabetes threshold against 4.0 per cent on placebo. Blood sugar levels went down, not up. Insulin sensitivity and insulin resistance were not measured directly by clamp in any of these programmes, so the improvement is inferred from glycaemia and weight rather than demonstrated at the tissue level.

Hypoglycemia is the risk that framing raises and it was not a prominent finding. GLP-1 receptor agonists stimulate insulin in a glucose-dependent way, which is why hypoglycemia with this class is largely confined to combination with insulin or a sulfonylurea rather than to monotherapy, and the dual agonists behaved the same way. What the glucagon arm has not demonstrated in humans is the thermogenic contribution it is credited with: no trial in this class has measured human energy expenditure by calorimetry, so the extra weight loss attributed to glucagon is inferred from the comparison rather than measured.

Side effects: the numbers that get left out

Gastrointestinal events dominate this class and the headline weight-loss figures are usually quoted without the tolerability figures that sit beside them in the same paper. Three numbers are worth carrying.

Mazdutide at 9 mg in GLORY-2: vomiting in 53.1 per cent, nausea in 46.9 per cent and diarrhea in 39.4 per cent of participants, with 2.9 per cent discontinuing for adverse events against none on placebo. High event rates, low discontinuation, which is the pattern of side effects that are unpleasant and manageable.

Survodutide tells a different story twice. In the phase two obesity trial, 75 per cent of participants on the drug had gastrointestinal events and only 60.4 per cent completed the 46-week treatment period at all. In phase three SYNCHRONIZE-2, gastrointestinal discontinuations ran at 18 per cent against 1.2 per cent on placebo, concentrated during dose escalation. That is the single most important number in the survodutide file and the one least likely to appear in a summary of it.

Retatrutide added a signal the GLP-1 class had not previously shown. In TRIUMPH-4, in participants with knee osteoarthritis, dysesthesia was reported in 8.8 per cent at 9 mg and 20.9 per cent at 12 mg against 0.7 per cent on placebo, with adverse-event discontinuations of 12.2 and 18.2 per cent against 4.0 per cent. Altered skin sensation at one in five on the top dose is a new finding for this class and belongs in any honest comparison of it.

What FDA has said about these three molecules being sold for research

This is the most directly relevant regulatory statement available for anything in this comparison, and it names three of the four molecules explicitly. FDA's page on unapproved GLP-1 drugs used for weight loss, content current as of 1 October 2026, states that retatrutide and cagrilintide cannot be used in compounding under federal law, being neither components of approved drugs nor found safe and effective for any condition. It then records that the agency has warned companies that illegally sold unapproved drugs containing semaglutide, tirzepatide, retatrutide, survodutide or mazdutide while labelling them for research purposes or not for human consumption, and selling them directly to consumers for human use with dosing instructions.

That is worth reading slowly, because it describes a specific pattern rather than the existence of research-grade material. What FDA objected to was the combination: research-use labelling on a product being marketed to consumers with instructions for taking it. The agency's position, stated in the same family of warning letters, is that research use only labelling does not determine intended use when the surrounding conduct points at human consumption. The same page reports Import Alert 66-80 covering GLP-1 active ingredients and, as of 31 May 2026, 990 adverse-event reports for compounded semaglutide and more than 730 for compounded tirzepatide.

The compounding route for the approved molecules closed earlier and has since closed further. FDA declared the tirzepatide shortage resolved in October 2024 and the semaglutide shortage resolved in February 2025, ending the enforcement discretion that had made compounded copies lawful, with the last deadlines passing in May 2025. Then in a Federal Register notice published 1 May 2026 the agency proposed not to add semaglutide, tirzepatide or liraglutide to the 503B bulks list at all, finding no basis to conclude the approved products are medically unsuitable and stating that a drug shortage does not constitute clinical need. Mazdutide, survodutide and retatrutide are not in that notice; they are covered by the alerts page instead.

On anti-doping, the position is narrower than many assume. No GLP-1 receptor agonist is on the WADA Prohibited List. Markers of semaglutide and tirzepatide are item 6 of the 2026 Monitoring Programme, in and out of competition, which carries no sanction but does mean prevalence data is being collected. WADA published the 2027 List on 8 October 2026 leaving that unchanged, citing insufficient evidence and a second study under way.

Frequently asked questions

References