Four ways in, each a real step — no obligation to climb them in order. The paid pilot is 100% creditable toward a subscription, so nothing you spend to evaluate is lost.
See the whole thesis, run the public alphabet test, read the L26 benchmark. Verify the gap yourself — no account.
Try it free →Score your own models on the Alphabet Test for the assigned languages (Bemba + Nyanja) — rates + L26, saved. The inventory itself begins at Pilot.
Request access → Most teams start hereA scoped, paid proof on your own held-out data, across your chosen languages and every domain. 100% credited toward a subscription.
Scope a pilot →License the whole living platform — every domain, every released language, the full consented corpus, and everything added while you're subscribed.
Talk to us →A separate, non-commercial track for grant, philanthropic, and impact funders — a curated Showroom view of the work and where it's going. Not a subscription, not for AI labs.
Pick your level above and tell us your model, target languages, and use case. Public needs nothing — just start.
Your request lands with our team. We confirm fit (and, for a pilot, a provisional scope) — usually within a couple of business days.
On approval we mint your access at the right tier and email you a one-time secure sign-in link — no password to set up.
Click the link and you land in your access: labs in their account with an API credential; funders in the Showroom.