
On December 22, Chile and Costa Rica signed a memorandum of understanding to collaborate on Latam-GPT, a regional large language model project for Latin America and the Caribbean. The agreement links Chile’s science ministry and Costa Rica’s MICITT, designating national AI centers to co-develop datasets, evaluation methods and governance standards for the shared model.
This article aggregates reporting from 1 news source. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Latam-GPT is part of the growing wave of ‘sovereign AI’ projects, but with a regional twist. By formalizing Costa Rica’s entry into a Chile‑led consortium, Latin America is trying to avoid a future where its languages, culture and legal norms are afterthoughts in US‑ or China‑centric models. A shared regional model, built with local datasets and governance, could give smaller states more leverage in negotiations with frontier labs and cloud providers, while seeding local ecosystems of fine‑tunes and applications.([portalinnova.cl](https://portalinnova.cl/chile-y-costa-rica-sellan-acuerdo-en-inteligencia-artificial/))
This won’t move global state‑of‑the‑art benchmarks, but it matters for diffusion: the more capable regional models exist, the less dependent local institutions are on a handful of US APIs. That, in turn, shapes where talent, startups and regulatory standards cluster. If Latam‑GPT succeeds in preserving regional linguistic and cultural nuance, it could become the default base model for governments, universities and SMEs across Spanish‑ and Portuguese‑speaking markets.
The risk is that under‑resourced public initiatives struggle to keep up with rapidly improving frontier models. Success will hinge on whether Latam‑GPT can plug into global open‑source ecosystems and attract sustained contributions, rather than becoming a one‑off science project.


