For AI labs & data strategists
Request evaluation access.
Evaluation lets you run the scored Alphabet Test on your own models, against the assigned evaluation languages — the deterministic measure of whether a model can reproduce a Bantu operating alphabet. You get the score, not the data: the full syllable inventories, audio, and downloads begin at the Pilot step, full depth, on your selected languages.
Evaluation
The scored Alphabet Test on the assigned languages (Bemba + Nyanja) — recall, precision, and your L26 score, with saved history you can re-run and track. Your own LLM or agent can run it too, by key. Rates only — no inventory, no audio, no downloads; the syllables stay licensed. Plus the free articles + public engines.
Pilot → Subscription
Pilot is where the data begins: full FSIs + every covered domain on your pilot languages; the Full Annual Subscription unlocks all 459 FSIs + every product line, full recording program, bulk export. License the program, not a dataset.
See exactly what Public · Evaluation · Pilot · Full Annual Subscription each unlock →
No public sign-up. Access is provisioned by BantuNomics. Already have a key? Log in.