Chile has entered the global artificial intelligence race with the launch of Latam-GPT, the first open-source AI language model trained specifically on Latin America’s cultures and linguistic nuances.

The initiative aims to reduce bias found in English-dominated AI systems and improve performance across regional contexts.

The project is led by the National Center of Artificial Intelligence (CENIA) and backed by more than 30 institutions across eight Latin American countries.

President Gabriel Boric described the model as a strategic step to ensure the region actively shapes the technological revolution rather than follows it.

Announced at the February 2025 AI Action Summit in Paris, Latam-GPT is designed as foundational infrastructure for future regional applications rather than a direct competitor to tools like ChatGPT or Gemini.

Developers say the model integrates large volumes of previously unavailable Latin American data to better reflect the region’s cultural realities.

Researchers collected more than eight terabytes of data, combining private partnerships and synthetic datasets to fill gaps in underrepresented areas.

The approach ensures the system captures linguistic and social particularities often overlooked in global AI training.

Initially, Latam-GPT will operate in Spanish and Portuguese, with plans to incorporate Indigenous languages in future updates. Experts say this expansion will further strengthen regional inclusivity and digital representation.

The launch also signals Latin America’s growing technical capacity to build sovereign AI systems. Analysts note that understanding and developing the technology locally strengthens regulatory oversight and long-term innovation strategy.

According to data from the University of Oxford, the United States, China, and the European Union host more than half of the world’s advanced AI data centers, while South America and Africa lag significantly behind.

Chile has responded by investing in talent development and expanding data infrastructure to close the gap.

Latam-GPT’s initial development was funded with approximately $550,000 from CENIA and the Development Bank of Latin America (CAF).

Future iterations will be trained on a $4.5 million supercomputer at the University of Tarapacá, with expanded training scheduled for 2026.