The evidence can say that both approved oral routes beat placebo by a wide margin in large randomized trials. It cannot say how they rank against each other, or against weekly injections, because no adequately powered published head-to-head trial has done that comparison. Everything currently ranking them is indirect, and indirect analysis carries limits worth understanding before acting on it.
The trials that actually exist
Oral semaglutide for weight management has two phase 3 readouts in the public record. OASIS 1, published in The Lancet in 2023, randomized 667 adults with overweight or obesity and without type 2 diabetes to a 50 mg daily dose or placebo for 68 weeks. Estimated mean body weight change was 15.1 percent against 2.4 percent, with gastrointestinal adverse events in 80 percent of the treated group and 46 percent on placebo. OASIS 4, published in the New England Journal of Medicine in 2025, tested the 25 mg dose in 307 participants and reported an estimated mean change of 13.6 percent against 2.2 percent at week 64, with gastrointestinal events in 74.0 percent versus 42.2 percent.
Orforglipron has a much larger dataset behind it. The phase 3 obesity trial randomized 3,127 adults without diabetes across three doses for 72 weeks, reporting mean weight change of 7.5, 8.4, and 11.2 percent against 2.1 percent on placebo. At the highest dose, 54.6 percent of participants lost 10 percent or more, 36.0 percent lost 15 percent or more, and 18.4 percent lost 20 percent or more. A second trial of 1,613 adults with type 2 diabetes and a body mass index of 27 or above, described in the prescribing information, reported smaller reductions of 5.1, 7.0, and 9.6 percent against 2.5 percent, a pattern familiar from injectable trials in populations with diabetes.
| Trial | Drug and route | Duration | Mean weight change | Placebo arm |
|---|---|---|---|---|
| OASIS 1 | Oral semaglutide 50 mg daily | 68 weeks | 15.1% | 2.4% |
| OASIS 4 | Oral semaglutide 25 mg daily | 64-week endpoint | 13.6% | 2.2% |
| Orforglipron trial 1 | Oral, three doses, no diabetes | 72 weeks | 7.5% to 11.2% | 2.1% |
| Orforglipron trial 2 | Oral, three doses, type 2 diabetes | 72 weeks | 5.1% to 9.6% | 2.5% |
| STEP 1 | Semaglutide 2.4 mg weekly injection | 68 weeks | 14.9% | 2.4% |
| SURMOUNT-1 | Tirzepatide weekly injection | 72 weeks | 15.0% to 20.9% | 3.1% |
Why lining up those rows is not a comparison
Each of those trials ran its own protocol, in its own population, against its own placebo response. Baseline weight differed, enrollment criteria differed, follow-up differed, and even the placebo arms landed between 2.1 and 3.1 percent. A gap of a percentage point or two between two rows sits comfortably inside that noise.
Several published indirect analyses have tried to close the gap, including a population-adjusted comparison of oral semaglutide against orforglipron and an indirect comparison of the 25 mg tablet against the 2.4 mg injection. These are legitimate methods, and better than eyeballing two abstracts. They remain statistical reconstructions built on assumptions about how populations match, and they do not carry the weight of randomization between the arms being compared.
Trial doses and label doses are not always the same number
Two traps sit in the numbers above. The OASIS 1 result came from a 50 mg daily dose that is not a marketed strength; the approved oral semaglutide product for weight management is the 25 mg tablet, and it produced a smaller mean reduction in its own trial. Reading OASIS 1 as a description of what the marketed pill does overstates the case.
The orforglipron trials carry the reverse issue. The published phase 3 paper reports doses of 6, 12, and 36 mg using the investigational formulation, while the prescribing information presents the same trials at the equivalent commercial strengths, topping out at 17.2 mg once daily. The milligram figures differ, the underlying trial does not, and anyone comparing dose numbers across the two documents will otherwise draw the wrong conclusion.
Where the evidence is genuinely thin
Cardiovascular outcomes are the clearest gap. The cardiovascular outcomes trial described in the semaglutide weight management label enrolled 17,604 adults with established cardiovascular disease and was conducted with the weekly injection. The tablet carries a cardiovascular risk reduction indication on that same label, but no separate cardiovascular outcomes trial of the tablet appears in the document. Two other indications, pediatric use from age 12 and noncirrhotic metabolic dysfunction-associated steatohepatitis, are assigned to the injection alone.
Durability data specific to oral products is also limited. What exists comes from the injectable record, where the STEP 1 trial extension found participants regained roughly two thirds of lost weight within a year of stopping, with cardiometabolic measures drifting back toward baseline. There is no reason to expect a tablet to behave differently, but that expectation is inference rather than evidence.
Real-world adherence is a third blind spot, and it bites hardest on the peptide tablets, whose trials enforced the fasting protocol that absorption depends on. Nobody has published a good estimate of how often people outside a trial dose correctly with water only and a 30-minute wait, so trial results and everyday results may separate further for oral semaglutide than for a weekly injection. Payment continuity sits in the same category: coverage for obesity medication is uneven, self-pay routes run from manufacturer channels such as NovoCare Pharmacy and LillyDirect to supervised telehealth programs from companies including Ro, Hims and Hers, and FormBlends, and no published trial measured what happens when someone stops because the price moved.
Compounded oral products have no comparable evidence at all
Compounded semaglutide sold as sublingual drops, troches, or lozenges is not FDA-approved, and compounded preparations are not reviewed by the agency for safety, effectiveness, or quality before marketing. No bioequivalence data supports treating them as equivalent to an approved tablet, and the trials above studied neither those formulations nor those delivery routes. Published pharmacovigilance work on compounded GLP-1 products describes the kind of signal that exists in place of trial evidence.
The contrast with the approved products is sharp on this point, since those come with both a trial record and a set of named suppliers a patient can actually vet. Self-pay routes run from LillyDirect and NovoCare Pharmacy to telehealth programs such as Ro, Henry Meds, and HealthRX, with HealthRX showing its GLP-1 medications and prices in the open. What no provider can supply is the head-to-head evidence described above, so a clear price and a documented clinical process are the things worth judging while the ranking data does not yet exist.
Frequently asked questions
Has any trial compared an oral GLP-1 directly against an injection?
Not in the published phase 3 record for weight management. Comparisons circulating between the 25 mg tablet and the 2.4 mg injection come from indirect treatment comparison and modeling studies, which reconstruct a comparison rather than randomize participants between the two arms.
Do the responder rates mean most people lose that much?
They describe proportions within one trial population. In the orforglipron phase 3 trial at the top dose, 54.6 percent reached 10 percent loss or more and 18.4 percent reached 20 percent or more, which also means a substantial minority did not reach 10 percent.
Is the diabetes evidence for oral semaglutide separate?
Yes. The glycemic control and cardiovascular indications for Rybelsus and Ozempic tablets sit on a different label from the weight management product, supported by their own trial program in adults with type 2 diabetes rather than by the obesity trials.
How much does the placebo arm matter when reading these results?
A great deal. Placebo groups in these trials received lifestyle intervention and structured follow-up and still lost 2 to 3 percent. Treatment effect is the difference between arms, not the headline figure, which is why cross-trial arithmetic on headline percentages misleads.
What would settle the ranking question?
An adequately powered randomized trial comparing the products head to head at their approved doses, with adherence measured rather than assumed. Until something like that is published, honest reporting describes each product against its own placebo and stops there.