How we grade peptide evidence
Every peptide gets an evidence band graded from public research registries — computed deterministically, refreshed nightly, and auditable back to the source. It grades the strength of the proof, never whether a peptide works.
We grade evidence strength, not efficacy or safety. A high band means the proof is strong, not that the compound is effective or safe to use. There is no LLM in the grade — it is computed from structured registry data, so there is nothing to hallucinate and every band traces back to a real record.
Approved as a medicine (FDA or a verified ex-US regulator), or Phase 3 controlled trials with published results.
Name-matched human trials with published or reported controlled evidence — real and ongoing, but not yet approved or definitive.
Human trials are registered, but none have reported controlled results yet. Most claims still rest on preclinical work.
Laboratory or animal data only — no name-matched human trials with reported results.
No indexed clinical or preclinical literature. Claims rest on user reports and marketing, not studies.
PubMed (NCBI)
Literature volume and publication types — randomized controlled trials, meta-analyses, and retractions.
ClinicalTrials.gov
US trial registry: phase, status, enrollment, primary outcomes, and whether results have been posted.
ISRCTN
International trial registry — catches registered trials outside the US that ClinicalTrials.gov misses.
openFDA
FDA drug labels (approval status) and FAERS adverse-event volume (shown as context for approved drugs only).
Health Canada DPD
Ex-US regulatory approval — so a compound approved outside the US is not wrongly capped below Established.
OpenAlex
Journal-venue tier and citation weight of the literature — grades the quality of the research, not just the count.
- Name-collision guard. We credit a study only when its title actually names the peptide — so "NAD" sunscreen labels don't inflate NAD+, and the "CardioGen-82" radiopharmaceutical doesn't make Cardiogen look FDA-approved.
- Registered ≠ proven. A band only reaches Promising when a trial has posted results or published controlled evidence exists — a merely registered trial keeps a compound at Early.
- Retraction check. We flag retracted publications in a compound's literature so a withdrawn paper never quietly props up a grade.
- Refreshed nightly. When a trial posts results or an approval changes, the band moves on its own — it is not a static bibliography someone has to remember to update.
- We don't make efficacy, dosing, or medical claims. This is research-use context, not medical advice.
- We don't grade whether a peptide is safe — evidence strength and safety are different questions.
- We don't let sponsorship or affiliate relationships touch the grade; it is computed from public registries.
- We deliberately under-grade rather than over-grade an obscure compound — it's better to be conservative than to overstate thin data.
See it on a profile
Every peptide profile shows its evidence band, the plain-English interpretation, the rubric inputs, and links to the key studies behind the grade.
Research-use only. PeptideBenchmark does not sell peptides and does not provide medical advice. Evidence bands describe the state of published research, which can change; they are not a recommendation to use any compound.