Every registered trial,
one question at a time
598,000 trials, their published outcomes and the FDA safety record behind every molecule — reachable from a script, a notebook or Claude, with a key of your own.
$10/month — about ten minutes of an analyst’s time. Cancel whenever. The documentation is public before you pay.
Rituximab in lymphoma · survival
- 0.74
- 0.23
- 0.51
- 0.75
- 1.10
- 1.08
The whole corpus
Trial search and aggregates, FDA drug and device safety, sponsors, targets and publications — the same endpoints the site itself calls.
See every endpointAnalysis at working volume
100 completion predictions, 100 pooled meta-analyses and 100 protocol benchmarks a month, where Free gives a daily taste of each.
How the numbers are computedBuilt for an assistant
Predictable paths, JSON responses, no session to manage, and errors that name the next step — so Claude can drive it without glue code.
Get the free skills/landscape/summary?target=TIGIT&phase=PHASE3
Sponsor ranking, phase split and completion rates in one response.
Claude skills for this work — free, no sign-up
Seven skills that teach Claude how to use this data properly: which endpoint answers which question, and — more importantly — when a number does not mean what it appears to mean. Take them whether or not you ever buy anything.
Map what has been tried before designing the next study, and read a safety profile the way a reviewer would.
Find who else is in your space and who is a plausible counterparty — with the NCT IDs to prove it.
Check a deck against the registry, and see what changed across your holdings this quarter.
All three build on /trialytic-api, the shared skill that covers endpoints, limits and the honest limits of the data.
Searching trials, molecules, devices and sponsors is free and needs no account. Pro is about volume and programmatic access.
| Capability | Free | Pro |
|---|---|---|
| Trial completion predictions | 3 / day | 100 / month |
| Meta-analyses | 2 / day | 100 / month |
| Protocol benchmarks | 2 / day | 100 / month |
| Exports | 5 / month · 100 rows | 500 / month · 1,000 rows |
| API requests | — | 1,000 / day |
| Monitored searches | — | Included |
These are the limits the server enforces. The API rows are documented in full, publicly, in the API reference, and your live remaining allowance is on your account page.
The things people
actually ask
Trials come from ClinicalTrials.gov and are refreshed on a schedule, with newer records merged from the registry's bulk JSON. The FDA layer — FAERS, MAUDE, labels, approvals, recalls — is rebuilt monthly.
The API answers 429 with the time your window frees up. It is a rolling 24-hour window rather than a midnight reset, so a job that starts late in the evening keeps its allowance. Nothing is charged for going over and nothing breaks.
So the service stays fast for everyone. Analysis and aggregates are deliberately generous — a summary across 50,000 trials is a single call — but the corpus is not downloadable in one request. Ask a question and the answer arrives in one response.
Any time, from your account page. Access runs to the end of the period you have already paid for rather than stopping that day, and the documentation stays public either way.
No. Searching trials, molecules, devices, sponsors and targets is free and needs no account, and the full API reference is public. Pro raises the ceilings and issues the key.
No. It is a research tool over public records. Reported safety counts reflect reporting volume, not incidence or causation, and pooled estimates describe registered results rather than clinical recommendations.
Ask the registry directly
$10 a month. Cancel whenever — access runs to the end of the period you paid for, and the skills and documentation stay free either way.