Answering the narrow version, because the broad one does not have a single answer. SURMOUNT-1 landed around 20.9% mean weight loss at 15mg over 72 weeks, against roughly 15% for semaglutide 2.4mg in STEP 1. Different trials, different populations, so the comparison is indicative rather than decisive — but SURPASS-2 was a genuine head-to-head and it pointed the same way.

anders_CPH said:

SURMOUNT-1 landed around 20.9% mean weight loss at 15mg over 72 weeks, against roughly 15% for semaglutide 2.4mg in STEP 1.

Agreed on the mechanism, with the caveat that the head-to-head used semaglutide 1mg, not 2.4mg. It is still the best direct evidence available, but it is not the comparison most people think they are citing.

That is the short version; the long version is somebody else's post.

alex_tucson said:

Fourteen months on tirzepatide, currently 10mg, and I stopped escalating because 10mg is doing the job and 15mg made me feel flat rather than full.

Can confirm the pattern alex_tucson describes. The GIP arm is doing real work rather than padding the label. GIP receptor agonism appears to improve adipose insulin sensitivity and lipid handling, and — counter-intuitively — GIP signalling in the CNS reduces nausea rather than adding to it, which is why tolerability at high total agonism is better than the GLP-1-only comparison would predict. SURPASS-2 is the cleanest head-to-head: tirzepatide beat semaglutide 1mg at every dose tier.

Adding the clinical framing, because it changes how the question reads.

Dose-response modeling for tirzepatide: Emax model fitting to the STEP/SURMOUNT dose-finding data shows:

Semaglutide: ED50 ≈ 0.6mg, Emax ≈ -18%, Hill coefficient ≈ 1.3
Tirzepatide: ED50 ≈ 6mg, Emax ≈ -25%, Hill coefficient ≈ 1.5

Clinical implication: most patients achieve >80% of maximal response by the mid-range dose (1.7mg sema, 10mg tirz). Going to the maximum dose provides diminishing returns — possibly not worth the additional side effect burden for some patients. Individualize dosing based on response vs tolerability.