Semaglutide vs tirzepatide vs liraglutide: how the three peptide APIs compare

Short answer: liraglutide and semaglutide are GLP-1 receptor agonists built on the 31-residue human GLP-1(7–37) backbone, each with a fatty acid at lysine 26 to bind albumin. Tirzepatide is a 39-amino-acid, GIP-based peptide with a C20 fatty di-acid at lysine 20 that activates both GIP and GLP-1 receptors. Those design choices explain their half-lives: about 13 hours, 1 week and 5 days.

This is a structural comparison for people who specify, buy or develop the APIs, not a clinical one, and it contains no dosing or treatment information. Semaglutide and tirzepatide are available from Winhycare Pharma for research use only (liraglutide on request as a custom peptide). They are compared here because buyers compare them constantly.

How do semaglutide, tirzepatide and liraglutide compare at a glance?

Attribute	Liraglutide	Semaglutide	Tirzepatide
Sequence basis	Human GLP-1(7–37)	Human GLP-1(7–37)	GIP sequence
Length	31 amino acids	31 amino acids	39 amino acids
Backbone changes	Lys34→Arg	Ala8→Aib, Lys34→Arg	Aib at positions 2 and 13; C-terminal amide
Lipid side chain	C16 fatty acid (palmitic acid) on a glutamic acid spacer	C18 fatty di-acid on a hydrophilic γGlu-2xOEG linker	C20 fatty di-acid (1,20-eicosanedioic acid) via a linker
Attachment site	Lys26	Lys26	Lys20
Receptor target(s)	GLP-1 receptor	GLP-1 receptor	GIP and GLP-1 receptors
Half-life (US label)	13 hours	About 1 week	About 5 days
Molecular formula	C172H265N43O51	C187H291N45O59	C225H348N48O68
Molecular weight (US label)	3751.2 Da	4113.58 g/mol	4813.53 Da
CAS number	204656-20-2	910463-68-2	2023788-19-2
First US approval	2010	2017	2022
First EU authorisation	30 June 2009	8 February 2018	15 September 2022

Sources: structure, molecular weight, half-life and initial US approval year from the FDA prescribing information for each reference product (semaglutide, tirzepatide, liraglutide); tirzepatide length from Coskun et al. (2018); semaglutide linker from Knudsen and Lau (2019); CAS numbers from PubChem (semaglutide, liraglutide, tirzepatide); EU dates from the EMA medicines report. The 31-residue length follows from the GLP-1(7–37) numbering, positions 7 to 37 inclusive.

What is the difference between GLP-1 and semaglutide?

Native GLP-1 is a hormone that barely survives in the bloodstream. The liraglutide prescribing information notes that GLP-1(7–37) “has a half-life of 1.5 to 2 minutes due to degradation by the ubiquitous endogenous enzymes, dipeptidyl peptidase 4 (DPP-4) and neutral endopeptidases (NEP).”

Semaglutide fixes both problems. Its discovery paper describes “two amino acid substitutions compared to human GLP-1 (Aib(8), Arg(34))” and derivatisation “at lysine 26” (Lau et al., 2015). FDA’s label explains the roles: the position 8 change provides “stabilization against degradation by the enzyme dipeptidyl-peptidase 4”, while the Lys26 modification with “a hydrophilic spacer and a C18 fatty di-acid” makes albumin binding “the main protraction mechanism.” The Arg34 change is described as “a minor modification… to ensure the attachment of only one fatty di-acid.” Without it, there would be two lysines to acylate.

So semaglutide is still recognisably GLP-1. It has the same receptor, a 31-residue backbone and two changed positions, plus one engineered side chain that changes how long it lasts.

How does liraglutide differ from semaglutide?

They are siblings from the same research programme, and the differences are small on paper but large in behaviour.

Liraglutide’s precursor is “engineered to be 97% homologous to native human GLP-1 by substituting arginine for lysine at position 34”, and the drug is made by “attaching a C-16 fatty acid (palmitic acid) with a glutamic acid spacer on the remaining lysine residue at position 26” (FDA, NDA 022341). Position 8 is unchanged.

Semaglutide makes three moves on that template:

1. Aib at position 8 for DPP-4 resistance.
2. A longer lipid, a C18 di-acid rather than a C16 mono-acid.
3. A longer, more hydrophilic linker, γGlu-2xOEG. Knudsen and Lau report that “a C18 di-acid together with a γGlu-2xOEG linker resulted in the highest albumin affinity” in the series they tested.

The trade-off shows up in receptor binding. Lau and colleagues measured semaglutide’s GLP-1 receptor affinity at 0.38 ± 0.06 nM, “three-fold decreased compared to liraglutide, whereas the albumin affinity was increased.” Lower affinity at the receptor, much longer exposure in the body. In the US labels, that is the difference between a 13-hour half-life and roughly a week.

How is tirzepatide different from both?

It starts from a different hormone. FDA’s label describes tirzepatide as “based on the GIP sequence”, containing “aminoisobutyric acid (Aib) in positions 2 and 13, a C-terminal amide, and Lys residue at position 20 that is attached to 1,20-eicosanedioic acid via a linker” (FDA, NDA 215866). The discovery paper calls it “a 39 amino acid linear peptide conjugated to a C20 fatty diacid moiety” (Coskun et al., 2018).

Its receptor profile is deliberately unbalanced. In binding studies its affinity was “comparable to native GIP for the GIPR and approximately 5-fold weaker than native GLP-1 for the GLP-1R”, and in cell signalling its potency was “similar to native GIP and approximately 13-fold weaker than GLP-1.” The same paper found it less potent at the GLP-1 receptor than semaglutide.

Protraction works the same way as for semaglutide. The label says the C20 fatty diacid “enables albumin binding and prolongs the half-life”, reports 99% binding to plasma albumin and an elimination half-life of about 5 days. Coskun and colleagues measured a mean of 116.7 hours in their early clinical study.

Why do their half-lives differ?

Two levers, used to different degrees:

Lever	Liraglutide	Semaglutide	Tirzepatide
Enzyme stability	Position 8 unchanged; label states it is stable against DPP-4 and NEP	Aib at position 8, for stability against DPP-4	Non-coded Aib at positions 2 and 13
Albumin binding through a lipid	C16 fatty acid; >98% plasma protein bound	C18 fatty di-acid; albumin binding is the “main protraction mechanism”	C20 fatty di-acid; 99% albumin bound
Half-life (US label)	13 hours	About 1 week	About 5 days

The table is drawn from the three FDA labels and the discovery papers cited above. Longer di-acid chains on suitable linkers raised albumin affinity in the semaglutide programme, and stability against DPP-4 and NEP removes the enzymes that clear native GLP-1 within minutes.

How do you tell the three APIs apart analytically?

Mass is the quickest check, provided you compare like with like. The FDA labels give average molecular weights. PubChem also lists monoisotopic masses, which is what high-resolution mass spectrometry usually reports:

Molecule	Average molecular weight (FDA label)	Monoisotopic mass (PubChem)
Liraglutide	3751.2 Da	3748.95 Da
Semaglutide	4113.58 g/mol	4111.12 Da
Tirzepatide	4813.53 Da	4810.52 Da

The tirzepatide discovery paper quotes a molecular weight of 4810.52 Da (Coskun et al., 2018), which matches the monoisotopic value rather than the label’s 4813.53 Da. Neither figure is wrong; they measure different things. A specification or certificate should say which one it uses.

On average mass, semaglutide is about 362 Da heavier than liraglutide and tirzepatide about 700 Da heavier than semaglutide, so a mix-up between the three is easy to catch by mass spectrometry. Mass alone still doesn’t prove the sequence. EMA’s synthetic peptides guideline recommends at least two orthogonal identity methods, chosen from mass, relative retention time, LC-MS, peptide mapping, amino acid analysis, bioactivity or NMR, that together “unambiguously confirm the sequence.” Our guide to reading a semaglutide API COA covers how those results appear on a certificate.

What do the structural differences mean for API manufacturing and specification?

For a buyer or developer, the comparison changes what the API specification and the impurity file have to cover.

• Length. A 39-residue peptide means more coupling steps than a 31-residue one in a stepwise synthesis, and more chances for deletion and insertion sequences. Our guide to SPPS vs LPPS covers route choices.
• Lipidation. All three carry a lipid side chain attached at a single lysine. EMA’s synthetic peptides guideline says conjugation adds complexity to characterisation and control, treats the unconjugated peptide as an intermediate that must be controlled, and names “the amount of the free unconjugated peptide and the free form of the conjugate moiety” as an additional quality attribute.
• Reference product route. FDA’s labels say the semaglutide backbone “is produced by yeast fermentation” and the liraglutide precursor by “expression of recombinant DNA in Saccharomyces cerevisiae.” The tirzepatide label does not describe the production route. A synthetic generic of a recombinant reference has to show its impurity profile compares with that reference.
• Regulatory comparison. FDA’s July 2026 Federal Register notice announced revised draft product-specific guidances for all three molecules (FR Doc. 2026-15285).

What is the approval and generic status of each?

All three are active ingredients of approved medicines in the US and the EU (dates in the first table). Generic status differs:

• Liraglutide. FDA’s Orange Book data files list approved ANDAs, the first dated 23 December 2024. In the EU, the EMA medicines report shows two liraglutide medicines from a second marketing authorisation holder authorised on 15 July 2026.
• Semaglutide. The Orange Book data files downloaded in October 2026 list no approved ANDA. The EU market-protection period for the first reference product runs to February 2028 at the earliest; patents and supplementary protection certificates run separately, as our guide to semaglutide patent expiry explains.
• Tirzepatide. No ANDA is listed; it remains with the originator in both regions. See our tirzepatide page for more background.

Winhycare Pharma supplies semaglutide API for pharmaceutical manufacturing and R&D. Semaglutide and tirzepatide are also available for research use only, and liraglutide can be discussed as a custom peptide, not for human or veterinary use; research-grade material is not GMP API and cannot be used in a medicinal product.