AIMARA LAB and CENIA Forge Alliance for a Foundational Large Language Model for Latin America

We are proud to announce a historic specific collaboration agreement between AIMARA LAB and the National Center for Artificial Intelligence of Chile (CENIA). This five-year alliance marks a pivotal moment for AI in our region, uniting two key institutions to tackle a monumental challenge: creating a foundational Large Language Model (LLM) for Latin America.

The project, named the Latam GPT "Gran Modelo del Lenguaje (LLM) Latinoamericano," is led by CENIA and aims to develop an open-source large language model that will serve as a cornerstone for technological and scientific advancement across the region.

A Collaborative Effort for Regional Advancement

This project is a testament to the power of collaboration. Both institutions will bring their unique strengths to the table to ensure the success of the Latin American LLM.

AIMARA LAB's Commitments include:

• - Supporting the collection of open data in various domains of Latin American Spanish.
• - Providing technical assistance during the training process of the Large Language Models.
• - Promoting the exchange of academic personnel, including professors and researchers, to foster shared knowledge.

CENIA's Commitments include:

• - Providing all necessary computational power and equipment for the training, testing, and alignment of the LLM.
• - Managing all legal aspects related to data governance, including licensing and availability.
• - Ensuring access to all necessary technical resources for the project's development.

Openness as a Core Principle

A fundamental pillar of this project is the commitment to openness. CENIA will guarantee that the license for all elements of the Latin American LLM—including the data, architecture, and final model weights—is as open as possible, aiming for standards like CC-BY 4.0 or MIT. This ensures that the resulting technology will be a public good, accessible to researchers, developers, and innovators throughout Latin America.

The final product will be a neural network trained on Spanish language data, including its architecture and final weights.

A New Chapter for AI in Latin America

We are honored to join forces with CENIA on this transformative initiative. This agreement is more than a technical project; it is a shared vision to build technological sovereignty and empower our regional scientific community. Together, we are laying the groundwork for a future where AI is developed by and for Latin America.