The marketplace for
AI training data.

Aribase helps companies license proprietary code, documents, audio, video, and business records to AI labs. Our specialists assess each asset and match it with relevant buyer demand.

Request an assessment Discuss your data Browse the catalog

The catalog at a glance

  • 18,371 code repositories
  • 31,832 data assets
  • 50,203 assets tracked

The catalog lists a selection of these, assessed and anonymized. Browse the catalog

Examples

Recently assessed

Real assets, described without naming the owner.

  • Structured & Financial

    Product catalog for books and other merchandise

    • Over 40 million products, with millions of book records.
    • Titles, categories, brands, formats, availability, and merchandising features.

  • Video

    Fixed-camera security footage

    • Recordings from fixed cameras covering the car park and surrounding areas of an airport.
    • Four separate terminals, giving multi-location perspectives.

  • Images

    Diagnostic radiology archive

    • About 150 million diagnostic radiology images in DICOM format.
    • MRI studies and other modalities such as X-ray and CT.
    • Images and reports on a unified radiology platform across multiple imaging sites.

Data categories

What you can license

We evaluate assets across seven categories, with specialists assessing their quality, rights, and fit for buyer requirements.

Evaluation

How we evaluate data

Our specialists review each submission against three criteria.

  • Provenance and rights

    Who created the data and what can be licensed.

  • Depth and context

    Useful history, metadata, documentation, and relationships.

  • Buyer fit

    Relevance to an actual training, evaluation, or research requirement.

Licensing potential depends on the asset's quality, rights, and current buyer demand.

Licensing

How licensing works

Four steps from a first submission to a written license agreement.

  1. Submit

    Tell us about the asset: what it is, how it was created, and what rights you hold.

  2. Assessment

    Our specialists review the asset's quality, rights, and fit with buyer demand. We aim to send an initial assessment within two business days.

  3. Buyer matching

    We introduce the asset to AI labs and model teams that have a matching requirement.

  4. License agreement

    The scope of rights is set out in a written license agreement, signed with Aribase or directly with the buyer, depending on the asset.

Buyers and experts

For buyers and experts

Browse the catalog

For buyers

Tell us what your model needs: modality, domain, scale, and rights. We look for assets that fit, from sources beyond the open web.

Request data

For experts

Paid AI training work for professionals: evaluate model outputs, create expert data, and lend your judgment to frontier systems. Remote and flexible.

Join the bench

Questions

Questions sellers ask

Do I keep ownership of my data?

Yes. Buyers receive defined rights to use the data for AI training and evaluation. Ownership stays with you, and you can keep using the data in your business.

Does Aribase buy my data?

No. Aribase licenses proprietary company data to AI labs and model teams. Depending on the asset, we license it from you and license it on to buyers, or we broker a license directly between you and the buyer. You keep ownership and can keep using your data.

Will anyone know my company is licensing?

No asset is listed publicly. Buyers see an asset only in anonymized form, without your company's name, and your company is named only with your consent, typically at contract time. Anonymized catalog entries are open to registered buyers. A buyer signs an NDA before receiving samples, data room access or prepared material.

How is Aribase paid?

There is no fee to submit an asset or receive an assessment. Aribase is paid only from licenses that close, a share agreed with you before any buyer sees the asset.

Is my license exclusive?

Only if you choose it and the buyer agrees. Exclusive means one buyer holds the licensed rights, while a non-exclusive license can go to several buyers. Every asset stays private and is never listed publicly, whether or not the license is exclusive. Pricing depends on the scope of rights, including any exclusivity.

How reliable is the first estimate?

It is an early indication based on the data categories, quantities and quality answers you give. Specialist review can move it up or down, and the final price is set in the written license agreement.

Insights

Notes on the market

Browse all insights

Start with an assessment.

We aim to send an initial assessment within two business days.

Request an assessment Discuss your data