Dihexa and Semax are sold in the same catalogues, often in the same "cognitive" bundle, and searched as if they were two options in one category. On the record, they are not. One is a registered medicine in another country, with a small and mostly Russian-language human literature that FDA read in full in 2026. The other has never been given to a person in a registered study, and the laboratory papers that established how it is supposed to work were withdrawn for fabricated data. This page sets the two records side by side, with today's counts, and states what neither contains. It does not rank them for use — neither has the evidence a ranking would need.
The records, side by side
| Semax | Dihexa | |
|---|---|---|
| What it is | Peptide, 7 amino acids: ACTH(4-7) fragment + Pro-Gly-Pro | Peptidomimetic from angiotensin IV (N-hexanoic-Tyr-Ile-(6)-aminohexanoic amide) |
| Formula and weight (PubChem) | C37H51N9O10S, 813.9 | C27H44N4O5, 504.7 |
| Registration | Registered in Russia, 0.1% and 1% nasal drops | None anywhere |
| Route in the human record | Intranasal only | No human record |
| People in studies FDA counted | 33–47 healthy adults, 69 adults with conditions, 451 children | 0 |
| ClinicalTrials.gov, 2026-10-11 | 0 studies | 0 (also 0 for PNB-0408) |
| PubMed records, 2026-10-11 | 232 | 19 (several unrelated) |
| FDA adverse-event reports | 1, through 2025-12-03 | 0 (openFDA, 2026-10-11) |
| Retracted papers | None found | 3, April 2025 |
| FDA compounding review | July 2026: staff advised against; committee voted 8–5 to recommend | Previously category 2, nomination withdrawn; FDA found no human exposure data by any route |
The molecular-weight row matters for anyone comparing milligram figures across vendor pages: a milligram of Dihexa contains about 1.6 times as many molecules as a milligram of Semax (813.9 ÷ 504.7 = 1.61, our arithmetic). Equal masses are not equal amounts, and neither compound has a dose-response study that would say what either amount does.
What Semax's record holds
FDA reviewed Semax for the July 2026 compounding meeting, nominated as a nasal spray or subcutaneous injection at 7,500 or 1,000 µg/mL for cerebral ischaemia, migraine and trigeminal neuralgia (FDA briefing document). Its reviewers found the human literature entirely intranasal — nothing by injection, the route half the nomination proposed. In eight references they counted 33–47 healthy adults, 69 adults with conditions including chronic ischaemic brain disease, facial and migraine pain and peptic ulcer, and 451 children with depression or tics.
Three of the eight were conference abstracts, several gave Semax alongside other treatments, and the doses as FDA printed them range from a single 0.5 mg/kg to "600 mg daily for 10 days" — a figure far above the others in the same table, which suggests a units error somewhere between the paper and the briefing. Only one reference reported on adverse events, and it reported none; the rest did not discuss them. FDA also flagged a paper describing possible anti-clotting properties, which raises a bleeding question for people on blood thinners that no study has addressed.
The registration is the strongest fact on Semax's side, and it is a Russian registration of nasal drops, not a finding by any regulator that has published its review. The detail of the human studies and the single US adverse-event report is on our Semax side-effects page; the figures the Russian label and studies use are on our Semax dosage page.
What Dihexa's record holds
Dihexa was developed at Washington State University as an orally active analogue of angiotensin IV, under the development code PNB-0408. Its marketed claim — that it binds hepatocyte growth factor, activates the c-Met receptor and builds new synapses — rests on a 2014 paper in the Journal of Pharmacology and Experimental Therapeutics. That paper and two companion papers were retracted in April 2025, after a university investigation found falsified or fabricated figure data, following a Notice of Concern the journal had carried since 2021. Our Dihexa dosage page gives the notice numbers and what a retraction does and does not establish.
What remains is animal work. PubMed returned 19 records for "dihexa" on 2026-10-11, two more than on 20 September: a 2025 rodent study of an HGF/MET "positive modulator" after repeated mild brain injury, and a 2026 orthopaedics review of therapeutic peptides. Several older hits are substring matches to unrelated chemistry. The clinical programme the laboratory's work led to, a different molecule called fosgonimeton, failed its primary endpoint in Alzheimer's disease in 2024.
There is no human dose, no human pharmacokinetic study and no human adverse-event record for Dihexa under any of its names.
Where the two differ in kind, not degree
- Evidence class. Semax has small, old, mostly unregistered human studies and a foreign registration. Dihexa has none. A side-by-side "benefits" table of the kind vendor pages publish puts a retracted rodent mechanism next to a registered medicine as if they were equivalent; they are not.
- Route. Semax's human record is nasal. Dihexa was designed to be taken by mouth. Vendor vials of both are commonly reconstituted for injection, a route with no human data for either.
- Regulatory attention. Semax has had a full FDA staff review, which recommended against it, and an advisory vote that went for it 8–5. Dihexa acetate was once nominated and placed in FDA's category 2 of substances that may present significant safety risks; the nomination was withdrawn, and FDA's note on it records no human exposure data by any route.
- The mechanism question. Activating HGF/c-Met is the opposite of what several approved cancer drugs do — they block it. That is not evidence Dihexa is harmful, but it is the reason a toxicologist would want long-term studies before anyone took it repeatedly, and none exist.
What neither record contains
- A registered, placebo-controlled trial measuring a cognitive outcome in healthy people.
- A head-to-head study of the two, of any kind.
- Any human data by subcutaneous injection.
- Any study longer than a few weeks with adverse events counted.
- Any independent assay of what research-labelled vials sold under either name contain — see how to read a COA.
How the newer vendor name Adamax relates to Semax is on our Adamax vs Semax page, and the wider category is covered in our guide to cognitive peptides. Peptide Lexicon's Dihexa and Semax entries describe each compound. Anyone weighing either for a memory, mood or neurological complaint should take the question to a clinician.
Sources and dates
- FDA. FDA Briefing Document, Pharmacy Compounding Advisory Committee, July 23–24, 2026: Semax-related bulk drug substances. https://www.fda.gov/media/193348/download — nomination, registration status, Table 3 (human references) and the FAERS search. Read 2026-10-11.
- PubChem: Semax CID 9811102 (C37H51N9O10S, 813.9); Dihexa CID 129010512 (C27H44N4O5, 504.7). Read 2026-10-11.
- ClinicalTrials.gov API v2: "dihexa", "PNB-0408" and "semax", 0 studies each, queried 2026-10-11.
- PubMed via NCBI E-utilities: "semax" 232 records, "dihexa" 19 records, queried 2026-10-11; newest dihexa records PMID 42825082 (Neurotrauma Reports, 2025) and PMID 41490200 (J Am Acad Orthop Surg Glob Res Rev, 2026).
- openFDA drug adverse event endpoint, medicinal product "dihexa": no matches, queried 2026-10-11.
- Retraction notices: Journal of Pharmacology and Experimental Therapeutics 392(4):103566, 103567 and 103568 (April 2025), as recorded on our Dihexa dosage page (read 2026-09-20).
- FDA, Certain Bulk Drug Substances for Use in Compounding That May Present Significant Safety Risks (category 2 page, content current as of 2026-04-22), dihexa acetate entry, as recorded on our cognitive peptides page.
- Committee vote: trade-press reports of the July 24, 2026 meeting, as recorded on our Adamax vs Semax page.