The Community Builder
Executive Director of Chile’s National Center of Artificial Intelligence, CENIA, who can show that a region builds its own foundation model.
He builds the regional model layer: an open-source language model trained on Latin American data and culture.
Latam-GPT launched on 10 February 2026 with support from more than 30 institutions in eight countries, built on $550,000 in funding.
Market Status reads Rodrigo Durán as The Community Builder. The friction this archetype resolves is that models trained mostly on English data reflect other regions poorly, and no single institution in Latin America could fix that alone.
He is executive director of CENIA, the centre that led a two-year regional effort backed by more than 30 institutions across eight countries. Latam-GPT acts as foundational infrastructure for regional applications and does not compete directly with consumer chatbots. Later versions are due to be trained on a supercomputer at the University of Tarapacá that costs about $4.5 million.
His stated aim is for Latin America to develop and understand this technology itself. He argues that regulation depends on that understanding, because a region cannot regulate what it does not understand.
Latam-GPT draws on private data obtained through partnerships and on synthetic data for underrepresented areas, so part of its source material is closed even though the model is open. An outside researcher expects it to be hard for the model to compete with larger corporations that have greater resources.
Against the journey, Durán is Building. The Data Economy’s micro-containers assume the data behind a model reflects the people who rely on it, and training data from Latin America was missing. Latam-GPT assembled more than eight terabytes of regional data, and its ethics team records where gaps remain, which is a provenance micro-container for the model layer. His work helps introduce that instrument for the Communications, Customer Success and Public Affairs functions in Spanish and Portuguese, with Indigenous languages planned for later stages.
Said Latam-GPT is trained on a proportion of Latin American data that did not exist online before, which allows more accurate performance for the region. Associated Press
Said the launch shows Latin America can develop and understand the technology, with implications for regulation. Associated Press
Entries are added as new public statements appear. Last updated 19 October 2026.
Funding the build, for readers who can direct capital toward the containerization.
Developing the micro-container that holds the gap, for readers who can make the technology.
Helping the micro-container reach scale, for readers who can set standards, convene consensus or move the political and regulatory conditions.
Note: The perspectives shared on GreenDeveX.com are the author's independent analytical interpretations of trust as a historical medium of exchange. This commentary is intended for conceptual exploration and does not constitute peer-reviewed academic research or commissioned institutional findings.