How long GLP-1 medicines take to work, and what plateaus look like in the trials

What the labels say about the first weeks

GLP-1 medicines are started low and stepped up. The labels list the schedule, and the trials used the same kind of schedule, so the early part of any trial curve includes weeks spent below the dose the label lists for ongoing use.

The long half-life matters too. Each step up in dose takes about a month to settle, which fits what the labels say about steady state. Stomach side effects cluster in the same window. In the three-year SURMOUNT-1 results for people with prediabetes, most gastrointestinal events came in the first 20 weeks, during dose increases. Side effects, week by week has more.

Label facts from DailyMed (Wegovy 6/2026, Zepbound 8/2026, Ozempic 5/2026). Half-life is how long the body takes to clear half of a dose. Steady state is when the drug level stops building from one dose to the next. The Wegovy label says semaglutide stays in the blood for about 5 to 7 weeks after the last injection. The Atlas lists the schedules to explain trial curves, not to guide use, and it gives no comparison between the two products' strengths.

What the trials show, point by point

Trials report averages. Two people on the same dose can sit far from the average, so these numbers describe groups.

Percentages are average changes in body weight from the start. Where a paper reports more than one way of counting, the Atlas uses the figure that includes everyone who was randomized. Company press releases often quote larger on-treatment figures, which should not be set beside these. The Atlas read the STEP 1 and SURMOUNT-1 trial results in the published papers and the 3-year SURMOUNT-1 and retatrutide results as abstracts.

People who start slowly

Not everyone loses weight at the same pace early on. In the SURMOUNT-1 tirzepatide group analyzed by Ard and colleagues, 82% had lost at least 5% by week 12 and 18% (278 people) had not.

Those 278 late responders did not stay put. By week 24, 194 of them (70%) had reached 5%, and by week 72, 250 (90%) had. Among the 250 who got there, the average time to reach 5% was 24.8 weeks. At week 72 their average loss was 11.0%, against 22.5% in the early responders.

A post hoc analysis of 1,545 tirzepatide-treated adults who took at least 75% of doses, so people who stopped early are not in it. The authors, several of them Eli Lilly employees, concluded that extending treatment well beyond 12 weeks may allow more people to reach a meaningful loss. The Atlas reports that as their conclusion. These are people who kept taking the medicine, and the numbers cannot say what will happen for any one person. The Ard paper cites a STEP 4 post hoc analysis of semaglutide 2.4 mg in which most people who were slow early on reached a clinically relevant loss with continued treatment. The Atlas did not retrieve that paper. A SURMOUNT-5 analysis found numerically more gastrointestinal and liver or gallbladder side effects in rapid responders on both medicines, with similar rates of finishing treatment.

When do the curves flatten?

The most direct look at plateaus on tirzepatide is a post hoc analysis by Horn and colleagues of two trials, SURMOUNT-1 and SURMOUNT-4. They defined a plateau as under 5% weight change over a 12-week stretch and over every 12-week stretch after it. They counted only people who took at least 75% of doses and had lost at least 5% by the end of the main period: 1,438 people in SURMOUNT-1 and 259 in SURMOUNT-4.

When the BMI groups are compared directly, plateaus began later in the higher starting-weight groups. By week 72, between 88% and 90% of each BMI group met the definition. In the paper's statistical models, though, BMI category and waist circumference were not linked to time to plateau, while dose, age and sex were. Weight kept edging down after the plateau began. In the SURMOUNT-1 group whose plateau began by week 24, average loss was 12.1% when the plateau started and 15.4% at week 72. The paper's model estimated that a plateau began 4.4 weeks later on 10 mg and 6.7 weeks later on 15 mg than on 5 mg.

Source: Horn and colleagues, Clinical Obesity 2025. Median times for class II and class III were significantly longer than for overweight, and the BMI-group figures are descriptive. The analysis is post hoc and exploratory, the authors say there is no agreed definition of a plateau, and most authors are Eli Lilly employees and shareholders. In SURMOUNT-1, 169 of the 1,438 people (11.8%) had not reached a plateau by week 72. Under this definition a plateau can still include slow loss, because weight can drop by under 5% in every 12-week stretch. Class labels follow the standard BMI bands.

What the authors said about the curve

• STEP 1: the paper says average weight did not reach its lowest point until week 60.
• STEP 5: the authors say loss plateaued after about week 60 and was maintained to week 104.
• SURMOUNT-1: the authors say 72 weeks was long enough to reach a plateau in the 5 mg group and near-plateaus in the 10 mg and 15 mg groups.
• Horn and colleagues: the authors say a plateau is often mistakenly seen as the end of the treatment's benefit or even as a loss of effect. They point to SURMOUNT-4, where continuing tirzepatide kept and added to weight loss compared with stopping.
• STEP 1 extension: after semaglutide stopped, the authors could not tell whether the slowing of regain was a plateau without longer follow-up. Weight regain after stopping Wegovy or Zepbound has the numbers.

Why plateaus happen: models and a mouse study

Three recent papers try to explain plateaus. Two are computer simulations and one is an animal study. None is a trial in people.

None of these papers shows that any plateau fix works in people.

Has anyone tested a fix?

No results exist yet. As of October 23, 2026, a search of ClinicalTrials.gov and PubMed turned up no published trial showing that an add-on medicine, a higher dose or a switch breaks a plateau in people who had one. It did turn up one registered trial built for that situation, described below. A search can miss things, so other trials may exist.

Advice online about breaking a plateau (more protein, strength training, calorie changes, dose changes or add-on medicines) has no results from plateaued people in the trials the Atlas found. The Atlas has separate pages on protecting muscle and the protein and strength reference, which cover what is known on those topics more generally.

• ENLIGHTEN-6 (NCT07392190, Eli Lilly, phase 3, 900 adults, recruiting since February 2026). Adults with obesity or overweight on a stable weekly incretin medicine, with body weight that changed by less than 5% at screening, get either eloralintide, an investigational once-weekly medicine, or placebo. The main weight measurement is at week 64. Type 2 diabetes is allowed but not required. The listing does not use the word plateau, and no results are posted. Primary completion is listed as June 2028.
• PLATEAU (NCT07446998, Veru Inc., phase 2, 239 people aged 65 or older with a BMI of 35 or higher). Despite the name, everyone starts semaglutide on day 1 and then gets either an experimental oral drug, enobosarm, or placebo. Weight is measured at 68 weeks. It does not enroll people who have plateaued. No results are posted, and primary completion is listed as October 2027.
• STEP UP (NCT05646706, Novo Nordisk). 1,407 adults were randomized to semaglutide 7.2 mg, 2.4 mg or placebo for 72 weeks, with no plateau requirement. Average loss was 18.7%, 15.6% and 3.9%. The 7.2 mg group lost 3.1 more points than the 2.4 mg group (95% CI 1.6 to 4.7). Stomach side effects were reported by 70.8%, 61.2% and 42.8%, and dysaesthesia (unusual skin sensations) by 22.9%, 6.0% and 0.5%. It shows a higher dose gave more loss in a general group. It does not show what happens in someone whose weight has stopped changing.
• Two registered switch trials, VESPER-SWITCH (Pfizer, NCT07845500) and MARITIME-SWITCH (Amgen, NCT07575399, open-label), move people from a weekly GLP-1 medicine to a different investigational one, with keeping weight off as the question. That is a different question from breaking a plateau.

Quick answers

How long do Wegovy and Zepbound take to start working?

In STEP 1, which used the Wegovy dose of semaglutide, weight loss was visible at the first assessment at week 4. In a SURMOUNT-1 analysis of tirzepatide, 82% had lost at least 5% by week 12. The Ozempic and Zepbound labels say the drugs reach steady levels after 4 to 5 weeks and 4 weeks, and the doses keep stepping up for months, so trial averages keep growing well past the first month.

How long does weight loss keep going?

In STEP 1 the average weight reached its lowest point at week 60. The STEP 5 authors say loss plateaued after about week 60 and held to week 104. The SURMOUNT-1 authors describe a plateau in the 5 mg group and near-plateaus at 10 mg and 15 mg by week 72. A phase 2 trial of the investigational drug retatrutide had lower averages at week 48 than at week 24 in every dose group.

What happened to people who had lost less than 5% by week 12?

In the SURMOUNT-1 analysis, 70% of those 278 people had reached 5% by week 24 and 90% by week 72. Their average loss at week 72 was 11.0%, against 22.5% in early responders. These were people who kept taking tirzepatide in a trial, and the numbers describe a group.

Are plateaus normal on GLP-1 medicines?

In the Horn analysis, 88% to 90% of eligible tirzepatide participants met a plateau definition of under 5% change over 12 weeks by week 72, though average weight kept edging down. Hall's simulation argues that plateaus are built into how the body balances energy. There is no agreed definition, so figures differ by paper.

Does a plateau mean the medicine stopped working?

The Horn authors say a plateau is often mistakenly seen as the end of the treatment's benefit or even as a loss of effect, and they point to SURMOUNT-4, where continuing tirzepatide kept and added to weight loss compared with stopping. That is the authors' reading. The Atlas cannot say what a plateau means for any one person.

Does a higher dose break a plateau?

The Atlas found no trial that tested that in people who had plateaued. STEP UP randomized adults to semaglutide 7.2 mg or 2.4 mg without requiring a plateau, and the higher dose averaged 3.1 more points of loss with more side effects. The Wegovy label (6/2026) lists a 7.2 mg dose for adults who tolerate 2.4 mg for at least 4 weeks when more weight reduction is clinically indicated, and it does not mention plateaus. None of this shows what happens in someone whose weight has stalled.

Do add-on medicines break a plateau?

No results exist yet. ENLIGHTEN-6, a Lilly phase 3 trial, is testing the investigational medicine eloralintide against placebo in adults whose weight is stable on a weekly incretin medicine, and its primary completion is listed as June 2028. The PLATEAU trial is named for the idea, but everyone starts semaglutide on day 1, and its primary completion is listed as October 2027. Neither has posted results.

Why does the curve start gently?

The labels start low and step up in stages. Wegovy's label starts at 0.25 mg and steps up every 4 weeks, with maintenance doses listed from week 17. Zepbound's starts at 2.5 mg, moves to 5 mg after 4 weeks, and lists further 2.5 mg steps after at least 4 weeks on a dose. The medicines also take about a month to settle at each step.

Questions to ask

• What dose schedule will I follow, and when do I reach the dose the label lists for ongoing use?
• What does this clinic count as a plateau, and how long does it wait before reviewing the plan?
• Which measures will we track besides the scale?
• What would you check first if my weight stopped changing?
• Is any part of my plan based on a trial, or is it common practice?
• Who decides when to review the plan, and what would trigger that?

What is still unknown

• No trial results show that any plateau fix works. ENLIGHTEN-6 is testing an investigational add-on in people with stable weight on a weekly incretin medicine, and it has no results yet.
• There is no agreed definition of a plateau. Horn and colleagues used under 5% over 12 weeks, and other papers use other rules.
• Plateau timing by dose is a model estimate (4.4 and 6.7 weeks later on 10 mg and 15 mg than on 5 mg).
• The early-response analysis the Atlas read is for tirzepatide. A STEP 4 analysis of semaglutide 2.4 mg is cited in it and was not read.
• Simulations show a possible mechanism, not what happens in a person. The Hall model rests on group averages.
• Averages hide wide differences. In the SURMOUNT-1 group whose plateau began by week 24, the average loss at week 72 was 15.4% with a standard deviation of 6.9 points.