Two obesity programs are being developed against the same receptor in opposite directions, and both of them work. Tirzepatide activates the GIP receptor alongside the GLP-1 receptor and produced a least-squares mean weight change of −20.2% (95% CI, −21.4 to −19.1) over 72 weeks in 751 adults with obesity and without diabetes.[1] Maridebart cafraglutide blocks the GIP receptor with an antagonist antibody while agonizing GLP-1, and produced −12.3% to −16.2% across its dose groups at 52 weeks in a 465-participant phase 2 obesity cohort, against −2.5% on placebo.[2] A receptor whose agonist and whose antagonist both strip double-digit percentages off body weight is not a solved mechanism, and the answer to what the second receptor contributes is genuinely unsettled.
The question is not academic, because the second receptor is the entire justification for a price step. Dual-agonist products are sold as a category above selective GLP-1 products, the difference in what tirzepatide takes off is real, and the explanation offered for it — a second incretin receptor doing additional work on appetite — is the part that has not been demonstrated. What follows is what has actually been measured, in which species, under which preparation.
On its own, GIP does nothing to appetite in humans
This is where the intuitive reading breaks first. Seventeen men with overweight or obesity were studied in a crossover design across four infusion days under matched glucose conditions: saline, GIP alone, GLP-1 alone, or both together, with energy intake at an ad libitum meal as the primary endpoint. GLP-1 cut intake to 2715 ± 409 kJ against 4483 ± 568 kJ on saline (p = 0.014). GIP alone produced 4062 ± 520 kJ — indistinguishable from saline (p = 0.590).[3]
Then the direction reverses. The combined infusion produced 3875 ± 451 kJ, which was not significantly different from saline (p = 0.364) and was significantly higher than GLP-1 alone (p = 0.039).[3] Adding GIP to GLP-1 did not potentiate the appetite effect in these men; the people receiving both hormones ate more than the people receiving GLP-1 by itself.
A second trial reached the same place from a different starting point. Twenty-two men with type 2 diabetes, already treated with metformin and a long-acting GLP-1 receptor agonist, received five-hour infusions of high-dose GIP or saline. Energy intake was 648 ± 74 kcal against 594 ± 55 kcal (P = 0.480), with no difference in appetite ratings or energy expenditure — while plasma glucagon and glucose both ran higher on GIP (P = 0.026 and P = 0.017).[4] Acute human infusion is a short instrument and cannot rule out an effect that needs months to appear. What it can do is establish that the simple additive story has been tested twice in people and failed twice.
The case for blocking it is older than the case for boosting it
The antagonist program did not come from nowhere. Mice lacking the GIP receptor were protected from obesity and insulin resistance on a high-fat diet, and double-homozygous mice lacking both the GIP receptor and leptin gained less weight and carried less fat than leptin-deficient mice alone; the knockout animals also ran a lower respiratory quotient, burning fat preferentially.[5] That was published in 2002, two decades before a GIP agonist was approved for weight.
Antibodies followed the genetics. A mouse anti-GIP-receptor antibody protected diet-induced obese mice against weight gain and reduced food intake; an anti-human antibody produced more pronounced weight loss in obese cynomolgus monkeys than the mouse version had in mice; and in both species, co-dosing the antagonist with a GLP-1 receptor agonist produced more weight loss than either alone. A conditional knockout in the same work excluded the pancreatic beta cell as the site of the effect.[6] Conjugating the antagonist antibody to GLP-1 peptides then produced greater weight loss in obese mice and monkeys than the antibody or a control conjugate alone, which the authors attribute to simultaneous receptor binding and rapid internalization amplifying endosomal signaling.[7] All of that is rodent and non-human primate. The human readout came later: a phase 1 study in participants with obesity showed dose-dependent weight loss that was maintained for up to 150 days after the last dose, a consequence of a molecule engineered for monthly administration.[8]
Tirzepatide is not semaglutide with GIP bolted on
The reason the head-to-head cannot settle the receptor question lives in the pharmacology. Receptor occupancy analysis at clinically effective doses found greater engagement of the GIP receptor than of the GLP-1 receptor — an imbalanced agonist, not a balanced one. At the GLP-1 receptor it is also biased: it favors cAMP generation over β-arrestin recruitment and drives weaker receptor internalization than native GLP-1 does. In primary islets, β-arrestin1 limited the insulin response to GLP-1 but not to GIP or to tirzepatide.[9]
So the molecule differs from a selective GLP-1 agonist in at least three ways at once: it hits a second receptor, it hits the first receptor with a different signaling profile, and it distributes its occupancy unevenly between them. Any weight difference between tirzepatide and semaglutide is the sum of all three, and no published analysis partitions it.
There is a structural reason no analysis can. Receptor occupancy is a property of a molecule at a dose, and the two drugs are not titrated to matched GLP-1 receptor exposure in any trial that has been run — they are titrated to maximum tolerated dose, which is a tolerability endpoint. Two compounds escalated to the point where their side effects stop a patient are being compared at whatever receptor coverage that happens to produce, and the coverage is different on each side. That is a perfectly good way to answer which product performs better in practice. It is not a way to isolate a receptor. The practical comparison for a buyer is in the molecule comparison; what follows is only about what the numbers can be used to claim.
What the head-to-head data can and cannot support
The obesity comparison is the strongest evidence in the class and the weakest possible evidence about GIP. In an open-label 72-week trial, tirzepatide at maximum tolerated dose produced −20.2% and semaglutide at maximum tolerated dose produced −13.7% (95% CI, −14.9 to −12.6), with waist circumference −18.4 cm against −13.0 cm (P < 0.001 for both).[1] That is a 6.5-percentage-point gap between two molecules. It is not a gap between GLP-1 and GLP-1-plus-GIP.
The diabetes head-to-head is weaker still for this purpose. Against semaglutide 1 mg — half the weight-management dose — tirzepatide at 5, 10 and 15 mg lowered HbA1c by 2.01, 2.24 and 2.30 percentage points against 1.86, and produced treatment differences in weight of −1.9, −3.6 and −5.5 kg.[10] A dose mismatch on one side of a comparison cannot be repaired by statistics, and that trial’s own page covers what it was powered to show.
The experiment that would answer the question is straightforward to describe and has not been run: one unchanged GLP-1 backbone, randomized with and without GIP agonism, at matched GLP-1 receptor exposure, in humans, over a year. Until it exists, every attribution of the tirzepatide margin to GIP is an inference from a comparison that confounded three variables.
The hypothesis that makes both directions true
One explanation reconciles the agonist and the antagonist without either being wrong: sustained agonism at the GIP receptor may desensitize and internalize it, so that a long-acting full agonist ends up producing something functionally close to blockade. The bispecific antagonist work reports rapid receptor internalization as part of its own proposed mechanism,[7] and the occupancy analysis of tirzepatide reports altered internalization at the GLP-1 receptor rather than the GIP one.[9] These are compatible observations, not a demonstration. No human experiment has measured GIP receptor availability during chronic treatment with either drug, so the reconciliation remains a hypothesis that happens to be convenient.
What none of this licenses at a checkout
Nothing here supports buying a second receptor as an upgrade. Maridebart cafraglutide is a phase 2 asset with no FDA approval and no product; cross-trial comparison of its 52-week figures against a 72-week tirzepatide trial with different estimands is not a ranking. The same caution applies to the triple agonists, where the phase 2 numbers are larger still and the approval is nonexistent, and to the GLP-1 and glucagon dual agonists covered in the survodutide and mazdutide article.
For anything actually purchasable, the dual-agonist question reduces to tirzepatide. Most sellers covered here dispense compounded tirzepatide or semaglutide, which is not FDA-approved and is not reviewed by the FDA for safety, efficacy or quality before it is dispensed — and a compounded preparation inherits none of the receptor pharmacology above unless the molecule in the vial is what the label says. The pricing side of that decision is the compounded tirzepatide board. The receptor argument is not a reason to pay more; the 72-week weight difference, with its open-label design attached, is the only argument the evidence actually supports.