Latin America Unites to Launch Culturally-Tuned AI Model 'Latam-GPT' This September

A coalition of over a dozen Latin American countries is set to launch Latam-GPT, the region’s first large-scale AI language model tailored to local cultures and languages. Backed by Chile’s National Center for Artificial Intelligence (CENIA) and more than 30 regional institutions, the open-source initiative aims to democratize artificial intelligence across Latin America.

Chilean Science Minister Aisen Etcheverry described the project as a "democratizing element for AI," envisioning its use in schools, hospitals, and public services where understanding regional context is critical. Unlike global models largely trained in English, Latam-GPT is designed to reduce bias and improve accuracy by focusing on the region’s linguistic diversity and cultural nuances.

Built on Meta’s Llama 3 architecture, Latam-GPT has been in development since January 2023. Training is conducted via a distributed network of Latin American computers, including those at the University of Tarapaca and cloud systems. The model is not intended as a direct competitor to consumer-facing tools like ChatGPT, but rather as foundational infrastructure for region-specific applications such as AI chatbots, personalized education tools, and virtual public service agents.

A standout feature of Latam-GPT is its commitment to Indigenous language preservation. A Rapa Nui translator has already been created, with plans to expand support to more Indigenous languages across Latin America.

Though the project currently lacks a fixed budget, regional development bank CAF and Amazon Web Services have offered support. CENIA Director Alvaro Soto remains optimistic that showcasing the model’s impact will draw further investment.

Set to launch in September 2025, Latam-GPT marks a major step toward inclusive, culturally aware AI development in the Global South.