Eli Lilly posted the full results of ACHIEVE-3 to ClinicalTrials.gov on 15 September 2026, and they are the first detailed head-to-head numbers between the two oral GLP-1 drugs. Over 52 weeks in 1,698 adults with type 2 diabetes, the higher dose of orforglipron lowered blood sugar further than the higher dose of oral semaglutide and took off nearly twice as much weight. This is a registry posting, not a peer-reviewed paper: no journal has reviewed the analysis, and the trial was open-label, so everyone involved knew which drug they were getting.
What the trial measured
ACHIEVE-3 randomised 1,698 adults into four groups: orforglipron at 12 mg or 36 mg once daily, or oral semaglutide at 7 mg or 14 mg once daily. Participants averaged 53.9 years old, 825 of 1,698 (48.6%) were women, and average HbA1c at the start was 8.29%. The primary question was whether orforglipron was no worse than oral semaglutide on HbA1c at week 52, using a non-inferiority margin of 0.3 percentage points.
It cleared that bar and went past it. HbA1c fell 2.16 percentage points on orforglipron 36 mg against 1.45 on oral semaglutide 14 mg — a difference of 0.71 points (95% CI 0.86 to 0.55 in favour of orforglipron). The lower orforglipron dose beat the lower semaglutide dose by 0.79 points, and beat the higher semaglutide dose by 0.46 points. By week 52, 85.4% of the orforglipron 36 mg group were under an HbA1c of 7.0%, against 66.1% on semaglutide 14 mg, a gap of 19.3 percentage points (95% CI 13.6 to 25.0).
Weight followed the same pattern. The orforglipron 36 mg group lost 9.2% of body weight against 5.3% on semaglutide 14 mg, a difference of 4.0 percentage points (95% CI 4.9 to 3.0), or 8.9 kg against 5.0 kg. Reaching at least 10% weight loss was roughly twice as common: 43.5% against 21.0%.
The side-effect ledger, and what the numbers do not settle
Gastrointestinal effects were more common on orforglipron, as they are across this drug class. On the 36 mg dose, 116 of 423 participants (27.4%) reported nausea against 90 of 425 (21.2%) on semaglutide 14 mg; vomiting ran 16.5% against 10.4%, and diarrhoea 23.4% against 18.1%. Serious adverse events were logged in 40 of 423 (9.5%) on orforglipron 36 mg against 24 of 425 (5.6%) on semaglutide 14 mg. Despite that, very few people quit over side effects — four on orforglipron 36 mg and five on semaglutide 14 mg — though overall completion was slightly lower on orforglipron (380 of 423, against 401 of 425). Two deaths occurred in the orforglipron arms and two in the semaglutide 7 mg arm.
The practical interest in orforglipron is that it is a small molecule rather than a peptide, so it does not carry the absorption constraints that shape how oral semaglutide is taken. A pill that works at least as well as the current oral option, without those constraints, is a different proposition for anyone who will not use an injection.
Several things are still open. A registry posting is a legal disclosure requirement, not a peer-reviewed publication: no independent reviewer has yet gone through the analysis, and the write-up will carry detail this posting does not. The trial was open-label, which matters most for the reported side effects and least for HbA1c. Everyone enrolled had type 2 diabetes with inadequate control, so these weight figures describe people taking the drug for blood sugar, not people taking it for obesity alone. The comparator was oral semaglutide, not the injectable version, which reaches higher exposures. And a 52-week trial of surrogate measures says nothing about heart attacks, strokes or deaths over years — nor about what regulators will make of any of it.
