A coalition of twelve Latin American nations is set to release Latam-GPT in September 2025, the first large language model tailored to the region’s cultural and linguistic diversity. Spearheaded by Chile’s National Center for Artificial Intelligence (CENIA), the open-source project brings together over 30 institutions across the continent. Rather than compete with ChatGPT, Latam-GPT is designed to support local services, such as schools, hospitals, and public administration. Built on Meta’s Llama 3 framework, the model promises to make AI more accessible and relevant in Latin America. A major goal is preserving Indigenous languages, starting with a translator for Rapa Nui.
Empowering Local Identity and Preserving Indigenous Languages
Latam-GPT aims to fill the gap left by English-trained global AI models, which often misunderstand regional dialects and context. According to Chilean Science Minister Aisen Etcheverry, this project could serve as a “democratizing force” by embedding Latin America’s cultural identity into AI systems. A significant feature is its focus on Indigenous language preservation. A Rapa Nui translator has already been developed, and similar efforts are planned for other Indigenous tongues. This move is vital for inclusive digital development, especially in education and public services, where native language access remains limited.
Rather than targeting general consumer use, Latam-GPT is built for public service applications. Like virtual assistants in government offices and AI systems in classrooms. These implementations will be more culturally relevant and linguistically accurate compared to foreign models. The AI is being trained using a distributed computing network. That includes Chile’s University of Tarapaca and cloud infrastructure provided by Amazon Web Services (AWS). Funding support has also come from the regional development bank CAF. The project remains without a fixed government budget, but CENIA officials believe that demonstrating Latam-GPT’s practical benefits will unlock broader financial and institutional backing across the continent.
Building Regional AI Infrastructure Through Global Collaboration
Latam-GPT is based on a distributed training infrastructure that ties on-premise university resources with scalable cloud services. The University of Tarapaca in Chile delivers compute capacity on a local scale, whereas AWS offers cloud-based assistance, which makes the business accessible and resilient locally. The model Llama 3 has been selected to be based on, due to its open-source license and flexibility. Such strategic positioning demonstrates the desire of this region to develop its AI capabilities without supporting the ecosystem of big tech only.
In Latam-GPT, the initial objective is not intended at commercial hegemony as is the case with Western AI superpowers, but rather, it will serve the greater benefit of the people as well as independence in its region. It is expected that the model is going to be used to develop the national-scale applications in healthcare, education, public administration, and emergency services. As the Director of CENIA, Alvaro Soto stated, the long-term vision of the initiative is the distribution of Latam-GPT in local schools and hospitals in order to enhance productivity and minimize interlingual barriers.
Although the project is yet to be funded under its national structure, its promoters are optimistic. That as soon as the positive impact it will have on society is seen, other investors will come to scale it. In addition to funding, open-source development guarantees the flow of additional contributions from academic organizations and developers. Which further strengthens the ecosystem. Latam-GPT is one of the most ambitious AI projects in the Global South. As it relies on the ideology of cultural specificity, digital sovereignty, and regional innovation.
A Vision for Sovereign AI in the Global South
Latam-GPT is a huge forward step that could lead to achieving digital sovereignty and AI equity in the Global South. Contrary to commercial AI models, which aim at the English-speaking market. This project focuses on cultural inclusion, linguistic diversity, and the delivery of public services. With the support of a coalition of Latin American institutions, cloud providers such as AWS, and development partners such as CAF. The model will create a vision in which AI matches the local realities in the region in the future. From supporting Indigenous language revitalization to streamlining government services. Latam-GPT aspires to empower communities across Latin America. Making AI not just more accessible but also more representative, responsible, and resonant with local values and needs.
About The Author
