Chile has taken a significant step in the global artificial intelligence race by launching Latam-GPT, the first open-source AI language model specifically trained on the diverse cultures and linguistic nuances of Latin America. This initiative aims to address existing linguistic biases prevalent in many AI models, which are predominantly trained on English data, and to ensure a more accurate, correct, and efficient performance when engaging with Latin American and Caribbean contexts.

Latam-GPT is the culmination of a two-year collaborative effort, spearheaded by the National Center of Artificial Intelligence of Chile (CENIA), and backed by over 30 institutions spanning eight Latin American countries. Chilean President Gabriel Boric emphasized the strategic urgency for Latin America and the Caribbean to play a proactive role in the technological revolution of AI, noting that Latam-GPT will be instrumental in integrating Latin American data and identity into the evolving AI landscape.

The project, which commenced in early 2023 and was announced at the February 2025 Artificial Intelligence Action Summit in Paris, is designed to serve as foundational infrastructure for future regional AI applications rather than directly competing with consumer-facing tools like ChatGPT or Google Gemini. Rodrigo Durán, executive director at CENIA, highlighted that Latam-GPT incorporates a significant proportion of Latin American data previously unavailable online and in other models. Gabriela Arriagada, a CENIA researcher and head of the project's ethics team, explained that developing Latam-GPT involved collecting more than eight terabytes of data, equivalent to millions of books, from private sources obtained through strategic partnerships, alongside synthetic data to cover underrepresented areas. This meticulous approach ensures the model reflects cultural realities and addresses gaps found in other models.

Luis Chiruzzo, an engineering professor at the University of the Republic in Uruguay, lauded Latam-GPT as a crucial milestone for the region. He noted that the inclusion of data capturing each country's particularities provides a vital assurance of inclusivity in the training process. Initially, the model will operate primarily in Spanish and Portuguese, with future plans to integrate Indigenous languages, further expanding its cultural reach.

According to Durán, the development of Latam-GPT signifies that Latin America now possesses the technical capacity to build its own AI models. This collaborative spirit not only fosters technological advancement but also carries important implications for AI regulation, as understanding the technology is fundamental to governing it effectively. While Chiruzzo acknowledged the challenge of competing with larger corporations boasting greater resources, he underscored Latam-GPT's importance as a significant step forward, enabling the region to establish its unique voice in the world of language models.

Globally, the race for AI leadership has prompted countries to re-evaluate their AI development policies. Data from Oxford University reveals that the United States, China, and the European Union host over half of the world's most powerful data centers for AI, in stark contrast to Africa and South America, which possess almost no AI hubs. In response, Chile has actively pursued expansion in the AI sector by attracting talent and building data centers, with President Boric stressing the necessity for the country to invest in AI to avoid falling behind in the future global landscape.

The initial development of Latam-GPT was funded with a modest $550,000 from CENIA's budget and the Development Bank of Latin America (CAF). The first version utilized Amazon Web Services' cloud and is set to launch at the end of February. Subsequent versions will be trained on a supercomputer at the University of Tarapacá in northern Chile, a significant investment of approximately $4.5 million, with training scheduled to commence in the first semester of 2026.