The marketplace for
AI training data
License datasets from 3,200 verified* sellers in minutes, not months. Search, preview, and stream straight into your training pipeline.
*Verification in progress. No credit card required.
Endurance Racing Pit Radio, 1998–2024
9,800 hours · Audio + text · Speaker-diarised
| timestamp | speaker | utterance |
|---|---|---|
| 03:12:44 | Engineer | Box this lap, box this lap. |
| 03:12:47 | Driver | Copy. Rears are gone. |
| 03:13:02 | Engineer | Fuel only, no tyres. |
| 03:13:05 | Driver | [redacted] |
# stream into training
import corpora
ds = corpora.load("pit-radio-1998")
for batch in ds.stream(size=512):
model.train(batch)Powering data teams at the world's most ambitious AI companies
Everything you need to buy and sell data at scale
From the first upload to the hundredth resale.
14,000+ datasets, one API
Text, audio, tabular, event streams. If it can be trained on, it's probably here.
40%
Week-over-week growth
Eleven weeks running. We stopped counting after that.
Model-ready in hours
Our pipeline cleans, labels and contextualises every upload, removing anything that looks too much like a name.
Pipeline: 82% de-identified
Get paid on every resale
Sellers earn a share every time their dataset is licensed. Again and again, in every jurisdiction we operate in.
“We went from zero to a production voice model in six weeks. I genuinely don't know where half this data came from, and honestly, that's the magic.”
DM Dana Moreau, Head of ML at Quanta Voice
Trending this week
| Dataset | Size | Format | Licence from | Status |
|---|---|---|---|---|
| Mayfair Domestic Correspondence, 1890–1912Digitised letters between butlers, footmen and housekeepers of London's finest households.… | 41,200 letters | Text | CHF 12,400 | Commercial |
| European Classic Car Auctions, chassis-levelHammer prices, estimates, matching-numbers status and full ownership chains for classic ca… | 2.1M rows | Tabular | CHF 38,000 | Commercial |
| Endurance Racing Pit Radio, 1998–2024Transcribed team radio from endurance races, speaker-diarised and timestamped. Profanity r… | 9,800 hours | Audio | CHF 54,500 | Commercial |
| DACH Doctoral Theses in LawFull text of doctoral theses in law from German, Austrian and Swiss universities, with cle… | 17,600 theses | Text | CHF 21,000 | Licence pending |
| Swiss Commercial Register, enrichedEvery commercial register filing since 2001, joined to professional networking profiles, h… | 6.4M records | Tabular | CHF 9,900 | Commercial |
| Consumer Consent Logs (EU)Cookie banner interactions with outcome labels: accept, reject, closed tab in frustration.… | 880M events | Event stream | CHF 74,000 | Consent status unknown |
| Grandmothers Reading Recipes AloudMultilingual voice notes of grandmothers reading family recipes. Collected via a recipe-sh… | 3,100 hours | Audio | CHF 16,750 | Commercial |
Your data is worth more than you think.
Most of it is already for sale. Make sure you get a cut.