Licensing & pricing
Licensing built for model training, not just for reading.
Annual licences in three paid tiers, plus a free sample. Commercial and Foundation / Lab include 12 months of monthly updates and a perpetual right to the snapshots you received; Research / Startup includes updates for the licence term. Licence agreement and quote on request.
We reply within one business day.
Last updated
Open sample
Evaluating fit, academic work, demos.
Free
upstream licences per record
- 1,233 records · 54 document histories (114 versions) · 2,017 typed edges, same schema as the licensed snapshot
- Includes the GDPR, AI Act, Digital Markets Act and Digital Services Act
- Each record carries its own upstream licence and attribution
- Arieta’s structure and metadata under CC BY 4.0
- Download now (.jsonl.gz), no sign-up
Research / Startup
Companies under about 50 staff, research labs.
from €7,000
per year
- Corpus, history and graph for the lanes you choose
- Internal R&D, evals and fine-tuning
- RAG behind your product for internal use and pilots
- Monthly updates for the licence term
Commercial
productionLegal-AI products in production.
from €18,000
per year
- Everything in Research
- Production RAG and fine-tuned models in your hosted services
- 12 months of monthly updates
- Perpetual right to delivered snapshots
Foundation / Lab
Foundation-model developers and AI labs.
from €50,000
per year · custom terms
- Everything in Commercial
- Pre-training and distribution in model weights
- Custom exports, cuts and delivery
- Perpetual right to delivered snapshots
Licence agreement on request: the Arieta Data Master Data Licence (English law, with tier schedules). We send it with your sample and quote; the final text is agreed with you before the first invoice. Licensed deliveries are prepared within 5–6 business days. Prices are indicative and exclude VAT.
-
01 · Reply
We reply within one business day, from a person, not a sales sequence. We confirm your jurisdictions, languages, dates and use case.
-
02 · Sample and licence
The free sample downloads today. A sample cut to your spec, with the schema, the licence agreement and a quote, follows within 5–6 business days.
-
03 · Order and delivery
A call if useful, then an order form for your tier. The licensed snapshot is delivered by signed download or to your bucket within 5–6 business days of the order.
Rights matrix
What each paid tier lets you do
Read across a row to compare tiers. The free sample is not in the matrix: each sample record follows its upstream licence, and Arieta’s structure and metadata in the sample are CC BY 4.0.
| Right | Research / Startup | Commercial | Foundation / Lab |
|---|---|---|---|
| Internal R&D, evaluation and benchmarkingExploration, analysis, internal eval sets. | Included | Included | Included |
| Fine-tuningSupervised fine-tuning or adapters on your models. | Internal use | Included | Included |
| RAG behind your own productIndex the data and retrieve it to ground answers. | Internal and pilots | Included | Included |
| Models served from your hosted servicesCustomer-facing features using models trained on the data. | Pilots only | Included | Included |
| Pre-trainingPre-training or continued pre-training of a base model. | Not included | Not included | Included |
| Release or sell model weightsDistributing weights trained on the data, including open weights. | Not included | Not included | Included |
| Redistribution of Arieta’s datasetPassing Arieta’s records, structure or graph to third parties. Upstream CC BY text stays redistributable under CC BY 4.0. | Not included | Not included | Not included |
| Perpetual right to delivered snapshotsKeep using what you received after the licence ends. | Delete on expiry* | Included | Included |
* Research / Startup: data is deleted on expiry, except CC BY and share-alike records, which stay under their open licences.
Need a right the matrix does not show? Foundation / Lab terms are written for your use.
Updates
Updates and the perpetual right, explained
01
Updates
Commercial and Foundation / Lab licences include 12 months of monthly delta snapshots: new and changed records (upserts), tombstones for withdrawn ones, and a fresh MANIFEST.json. Research / Startup licences include updates for the licence term.
02
Perpetual right (Commercial, Foundation / Lab)
Every snapshot delivered during a Commercial or Foundation / Lab licence stays licensed to you after the year ends, under the same tier rights. Tombstone obligations continue. Research / Startup data is deleted on expiry, except CC BY and share-alike records.
03
Renewal
Renewing for another 12 months of updates costs about 40–50% of the first year.
- base
- Δ1
- Δ2
- Δ3
- Δ4
- Δ5
- Δ6
- Δ7
- Δ8
- Δ9
- Δ10
- Δ11
- keep
Commercial example: base snapshot, then eleven monthly deltas during the licence year; after the year you keep everything delivered.
Key terms
Obligations that come with the data
The points the licence agreement is built around. They protect upstream publishers, the people named in court records, and you.
- No redistribution of Arieta’s dataset
- Licensees may not pass the dataset, or a substantial part of it, to third parties as a dataset, API or embedding dump. This covers Arieta’s structure, metadata, version history and citation graph, and lanes that are not under an open licence. Verbatim text from CC BY lanes stays under CC BY 4.0, which the licence does not restrict; those records are marked in every shard.
- Tombstone compliance
- When an upstream source withdraws a document, the next monthly delta carries a tombstone. Licensees must delete tombstoned records from their copies of the dataset, indexes and caches.
- Upstream attribution passes through
- Every record carries its source, licence and attribution, and each snapshot ships LICENCES.md. Where you publish material from the dataset, keep the attribution the upstream licence requires.
- Statutory restrictions pass through
- Some restrictions come from law, not from a licence. In France, reusing the identity data of judges and court clerks to evaluate, analyse, compare or predict their professional practice is a criminal offence (art. L111-13 Code de l’organisation judiciaire). The licence passes such restrictions on.
- Not the authentic text
- The dataset is a structured derivative of official publications. Only each official journal or gazette is legally authentic; consolidated texts are for information in most systems.
- Share-alike stays separate
- Italian Corte costituzionale decisions are CC BY-SA 3.0. They are delivered masked like the rest of national case law, as a separate CC BY-SA 3.0 file, never inside the main shards.
Tombstones: what they are and how withdrawals flow.
Procurement
The checklist, up front
For your procurement team
- Contracting entity
- Arieta Labs Ltd, England & Wales, company no. 16629785
- Governing law
- England and Wales
- Data protection
- Controller-to-controller data-protection terms in the licence agreement, sent on request
- Delivery
- Parquet (zstd) and gzip JSONL with MANIFEST.json (SHA-256 per shard), DATASHEET.md and LICENCES.md; signed download or your bucket, within 5–6 business days
- Updates
- Monthly deltas; upstream withdrawals are reflected as tombstones in the next delta
- Personal data
- No raw upstream HTML; national case law delivered with personal data masked
- Website
- No cookies; cookie-free Cloudflare Web Analytics only
- Certifications
- None claimed
FAQ
Licensing questions
Can I see the licence before I buy?
Yes. The licence agreement and your tier schedule are sent on request, with your sample and quote. The master licence is under legal review, and the final text is agreed with you before the first invoice; prices on this page are indicative.
Can we evaluate CJEU case law before licensing?
Yes. A CJEU case-law excerpt is available under an evaluation licence on request, prepared within 5–6 business days. Case law is not in the open sample, which is legislation only. Email us with the subject “Arieta Data: CJEU evaluation excerpt”. National case law is delivered with personal data masked; tell us which courts you need.
Can I pre-train a foundation model on the data?
Yes, on the Foundation / Lab tier, which covers pre-training and distribution in model weights. Research and Commercial tiers cover fine-tuning, RAG and evals.
What happens when my licence ends?
Commercial and Foundation / Lab licensees keep a perpetual right to every snapshot delivered during the licence, under their tier’s rights, and stop receiving updates. Research / Startup licensees delete the data on expiry, except CC BY and share-alike records. Tombstone obligations continue in every case.
What is a tombstone and what do I have to do?
A tombstone marks a record its upstream source withdrew, often for personal-data reasons. When a delta contains tombstones, you must delete those records from your copies of the dataset.
Can I buy only some jurisdictions?
Yes. Licences can be cut per jurisdiction, language, date range or document type, prepared within 5–6 business days. Research licences are priced by the lanes you choose; Commercial and Lab licences usually include all released lanes, and the order form lists them.
Can I redistribute the data?
Not Arieta’s dataset: its structure, metadata, history and graph stay under the licence. Verbatim text from CC BY lanes stays redistributable under CC BY 4.0, and each record says which licence applies.
Do you hold security or compliance certifications?
No. We do not claim any certification. We document how the data is produced, what is excluded and why, on the Provenance page.
Nothing on this page is legal advice. The licence agreement governs.
Get the licence agreement and a quote.
Tell us your use case, team size and jurisdictions. We send the licence agreement, your tier schedule and a sample cut for your lanes within 5–6 business days.
- We reply within one business day
- Custom cuts in 5–6 business days
- Free sample, no sign-up