Skip to content
Press kit

Brand & media resources

Logos, assets, and company information for journalists and partners.

What is Somast

Somast, short for Somali Science and Technology, is an open Somali-AI research lab. We make it easy for developers, researchers, and learners to build with language technology that understands Somali: open datasets, foundation models, and a free, hands-on data-science and ML training path, all released in the open.

Our story

Why we started

Somali is spoken by more than 20 million people, yet it holds an almost invisible share of the data that trains modern AI, a fraction of a percent of what English alone contributes. Every model, tokenizer, and benchmark we tried to use for Somali either didn't exist or quietly failed on the language's morphology and dialects.

So in 2023 we started building the missing layer ourselves: collecting and annotating text and speech, publishing it under open licenses, and training the first Somali-native benchmarks and models on top of it. Alongside the research, we opened a free Data Science & ML bootcamp so the people who will build Somali AI next could learn the skills to do it themselves.

Three years in, that combination (open data, open models, open education) has grown into a corpus of 1.1B+ tokens, 320+ hours of speech, a handful of Somali-native models, and 200+ alumni from our cohorts. We're still at the beginning of closing the gap.

Our mission

Open AI infrastructure for the Somali language, built with the community.

We believe a language shouldn't need a trillion-dollar company behind it to be understood by AI. Every dataset, model, and benchmark we publish is meant to lower that bar for the next person building in Somali.

Company details
Company name
Somast
Legal name
Somast LLC
Name stands for
Somali Science and Technology
Founded
2023
Headquarters
Mogadishu, Somalia
Website
somast.so
Brand blurb

One-liner

Open AI infrastructure for the Somali language.

Short (25 words)

Somast is an open Somali-AI research lab building datasets, foundation models, and a free data-science training path, all released in the open.

Long (100 words)

Somast, short for Somali Science and Technology, is an open research lab building the AI infrastructure the Somali language has never had. Instead of leaving Somali as an afterthought in models trained mostly on English, Somast collects and publishes open datasets, text and speech, and trains Somali-native tokenizers, models, and benchmarks on top of them, all under permissive open licenses. Alongside the research, Somast runs a free, hands-on Data Science & Machine Learning bootcamp so Somali engineers and researchers can build on that foundation themselves. Everything is public: the data, the model cards, the benchmark results, and the curriculum.

Key metrics
3
Open datasets
320h
Speech hours
1.1B+
Corpus tokens
6
Models in preview
48+
Contributors
973+
GitHub stars
Why people work with us
Fully open datasets
Somali-native models & tokenizer
Reproducible benchmarks
Free, structured bootcamp
Community-reviewed research
Built with Somali speakers
Brand assets

Logo package

SVG format, works on light or dark backgrounds.

Somast logo

Icon mark

Compact monogram for favicons, avatars, and small tiles.

Somast icon

Brand colors

Deep Teal accent on a calm, neutral base.

Deep Teal#117B72
Midnight Teal#0B5F59
Near Black#171918
Soft Surface#F4F6F4

Logo usage

Keep clear space around the mark of at least the logo's own height. Use on white, soft neutral, or near-black backgrounds.

Use the SVG wherever possible so it stays crisp at any size.

Don't recolor, recreate, add effects to, or stretch/skew the mark.

Don't place it on busy photography or low-contrast backgrounds.

Media inquiries

For press inquiries, interviews, or additional information, our team is glad to help.