Small, task-specific open models
Olaverse is an open-source multilingual AI infrastructure toolkit for building NLP, speech, retrieval, and language systems for underrepresented languages — built one focused model at a time.
25+ languages · Yorùbá · Igbo · Hausa · Swahili · Vietnamese · Turkish · Polish — and beyond
Featured models
/ 6 MODELSA cross-section of the stack, Stable and Labs alike — identify the language, restore its tone marks, generate from it, reach for the generalist LLM, then a peek at the vision line.
Start in seconds
Open weights, straight from Hugging Face, with the tools you already use. No API keys, no sign-ups, no rate limits.
from transformers import pipeline, AutoTokenizer, T5ForConditionalGeneration # 1 — Which language is this? lid = pipeline("text-classification", model="olaverse/lid-neural-25.2") lid("Kedu ka i mere?") >> [{'label': 'ibo', 'score': 0.998}] # 2 — Restore its tone marks (tag the target language) tok = AutoTokenizer.from_pretrained("olaverse/diacnet-1.0") model = T5ForConditionalGeneration.from_pretrained("olaverse/diacnet-1.0") out = model.generate(**tok("<yor> se eranko naa si gbo o?", return_tensors="pt")) tok.decode(out[0], skip_special_tokens=True) >> "ṣé ẹranko náà sì gbọ́ ọ?"
The full stack
/ 9 FAMILIESOur generalist LLM line, the task specialists, and the research lineage behind them — every model grouped by what it's built to do.
From the Blog
LATESTA diacritic model that cannot corrupt your text
How we cut Yorùbá diacritic error by 58%, by changing the shape of the problem, and by noticing what our training corpus was quietly teaching us to do wrong.
Introducing DiacNet-1.1: Multilingual Diacritic Restoration for 10 Languages
Pretrained From Scratch for Naija: Introducing mist-encoder-base-ng
The Small Model Behind a Better Chat List: Introducing mist-tg-0.3b
Release notes land on Hugging Face first.
Follow olaverse to catch every drop.
Build with our models
Every model is open-weight and free forever. Fine-tune it, deploy it, or drop it straight into your product. If you build something for underrepresented languages, tell us about it.