Renaissance Labs

Atlas / Europe

Academic lab · Bonn, Germany

Polygl0tTucano

A University of Bonn project building open foundation models for low-resource languages, best known for the Tucano Portuguese models and the GigaVerbo corpus.

language modelslow-resource languagesdatasetsevaluation

Bonn, Germany

Founded
2026
Type
Academic lab
Based in
Bonn, Germany
Key papers
2
Featured projects
7

About

Polygl0t, also referred to as Project Polyglot, develops efficient language models for low-resource languages. It is funded by the German BMBF and the North Rhine-Westphalia ministry as part of TRA Sustainable Futures at the University of Bonn, and trains on Bonn's Marvin cluster and Jülich's JUPITER Booster.

Its Tucano models, first described in 2024, were trained on GigaVerbo, a corpus of about 200 billion deduplicated Portuguese tokens. The follow-up Tucano 2 paper of March 2026 introduces GigaVerbo-v2, with about 320 billion tokens, and models from roughly 0.5B to 3.7B parameters.

The group publishes weights, datasets and training code, including an LLM foundry repository and a fork of the evaluation harness.

Mission and ambitions

Polygl0t aims to close the linguistic divide in NLP by making capable open models for languages that large developers serve poorly.

Known for

  1. 01

    Tucano Portuguese models

  2. 02

    GigaVerbo corpus

  3. 03

    Open training code

  4. 04

    Low-resource language models

Milestones

  1. 2024

    Tucano and GigaVerbo

    Released Portuguese decoder models and a 200B-token corpus.

  2. 2025

    Patterns publication

    The Tucano work was published in the journal Patterns.

  3. 2026

    Tucano 2

    Released a second model family with GigaVerbo-v2 and post-training datasets.

  4. 2026

    Workshop materials

    Published workshop slides and exercises on GitHub.

Open source

Research papers

Notable releases

Openness

Weights (Apache 2.0 on the instruct model checked), datasets and code are public on Hugging Face and GitHub. Brazilian focus but based in Germany.

People

Listed from the lab's own pages and public profiles.

Contact and community

Contact details and links are on the lab's website (https://github.com/Polygl0t); code questions go through issues on their GitHub organisation.

Sources

Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.

Labs nearby