Start with the sample. Then the tier that fits the job.
One tier you can buy outright, everything above it by agreement. In every case the models you train and their outputs are yours, deployable without further fee; the corpus itself is not redistributable.
Sample
- 17 items (
alfe-0001–0017) - Full schema and rationales
- Evaluation only, no redistribution
Pilot set
- 505 items (
alfe-0001–0505) - Every task type, register, domain and all fifteen error types
- JSONL, JSON review copy, build notes, schema card, licence certificate
- Non-exclusive, perpetual, internal use
Scaled, full & exclusive
- 2,000–5,000 items shaped to your taxonomy
- The full corpus: 3,009 items
- Exclusivity by field or language pair, or a bespoke build
- Annual refresh at 25% of the licence fee
Every tier.
| Tier | Contents | How it is bought | Price |
|---|---|---|---|
| Sample | 17 items, evaluation only | Form or Hugging Face | Free |
| Pilot set | 505 items, JSONL + JSON, build notes, licence certificate | Self-serve, click-through licence | $4,900 |
| Scaled | 2,000–5,000 items, shaped to your taxonomy, dialect mix and format | Quote, DLA | from $15,000 |
| Full corpus | All 3,009 items across 74 thematic chunks | Quote, DLA | Quote |
| Exclusive or bespoke | Exclusivity by field, or an authored build to your specification | Quote, DLA | Quote |
| Kept-current refresh | Annual update as statutes, AAOIFI standards and case law move | Recurring, licensees only | 25% / yr |
Prices in USD. The pilot set is fixed-price; tiers above it are scoped and priced by agreement. VAT applied where required.
Request a quote.
We reply within one business day with a one-page proposal: scope, item count, format, price, delivery date and licence type.
The models are yours.
Permitted use covers training, fine-tuning, evaluation, benchmarking and alignment. Models you train and their outputs belong to you and may be deployed and distributed without further fee. What is not permitted is redistributing the corpus itself, or using it to build a dataset that competes with it.
The pilot set is sold under a click-through licence you can read in full before buying. Larger tiers run under the full Data Licence Agreement, executed by e-signature.