Six worked pipelines, each built only on models we mark Stable — maintained, benchmarked, and recommended for production. Real code, published numbers, and the limitations spelled out.
from olaverse.nlp import Diacritizer, LIDNeural25
lid = LIDNeural25(variant="questions")
lid.load()
lid.predict("Kedu ka i mere?") # → 'ibo'
Diacritizer(model="auto").restore("Ojo lo si oja lana")
# → 'Òjó lọ sí ọjà lana'Every pipeline below runs on the same SDK — read the docs
Restore diacritics and tone marks on text that was typed, scraped, or keyed in without them — across 10 languages, from Yorùbá and Igbo to Vietnamese and Polish.
Detect the language of support tickets, chat messages, queries, and documents across 25 languages — including the African languages most detectors get wrong.
Generate natural, search-style questions from any passage across 25 languages — for quizzes, reading comprehension, help-centre FAQs, and assessment banks.
Generate a short, human title from a chat message or document in ~300M parameters — small enough to run on the same box as your app.
Restore the marks your documents lost, then match on an accent-folded key — so a user typing "oko" finds "ọkọ", and still sees it written properly.
Generate verified (query, passage) pairs from your own corpus, in 25 languages, and fine-tune retrievers and rerankers on data that matches your domain.
Not seeing yours? These are the pipelines we can point at published benchmarks for. Anything adjacent — a different language, a different task, your own data — we build to order.
Every model here is open-weight, so you can build any of it yourself today. If you would rather we adapted it to your data and shipped it, that is what Olaverse for Business does.