Venture companies in South and Central America get a jolt of life with a fast-evolving large language model called Latam-GPT. The project complements popular models, but with dynamic regional perspective. Importantly, AI systems often represent human biases in training data. As a sign of a complex worldview, there are more than 500 distinct indigenous languages in this region, as well as Spanish and Portuguese.
Latam-GPT sprouted from a Chilean policy initiative that created the National Center for Artificial Intelligence, commonly known by the Spanish acronym CENIA. With training data from 20 countries, including those in South and Central America, the Caribbean, and Spain, the model now has more than 50 billion parameters with context pulled from 2.7 million documents. In its current form, analysts suggest that Latam-GPT is broadly comparable to the GPT-3.5 model released by OpenAI at the end of 2022.
Is Latam-GPT a duplicative effort at a time when large language models are increasingly widespread? Processing training data in a region where structured information can be evasive is a Herculean task. While Brazil and other developing economies are deep into digitalization, Ecuador, Bolivia, and Venezuela are not. There is also a seemingly endless array of indigenous languages, including Guarani, Rapunati, and Mapuche. In the race to commercialization, the largest systems based in mature economies have likely overlooked this enormous potential.
Because Latam-GPT is an open-source model, there are myriad use cases that will benefit from its roll out. Companies with a wide social fabric, such as healthcare and education, may be among the first to gain traction from this culturally-responsive effort. Other applications might include credit scoring in banking or behavior analysis in retail commerce, given market-by-market features.
With an insatiable demand for computing power, further evolution of Latam-GPT will depend on the backbone provided by the supercomputer center at the University of Tarapacá in northern Chile. The national government is investing some $10 million in the facility, helping to affirm Chilean leadership in the field. The Latin American Artificial Intelligence Index, a public-sector assessment of AI development and readiness, ranks Chile first in the region, followed by Brazil and Uruguay. ■
Learn more at Wired
© 2025 Cranganore Inc. All rights reserved.
Unauthorized use and/or duplication of any material on this site without written permission is prohibited.
Image Credit: Khamz at Adobe Stock.