Follow the timeline
of our AI Model
Project BigBox
Follow the timeline
of our AI Model
Project BigBox
AI Research & Development
building the Future of Artificial Intelligence in Latin America
In MAGUS AI Suite, we believe that the future of artificial intelligence should also be built in Latin America. For this reason, we have begun the development of our own language model (LLM), a private research project driven by PCmX Public Technologies and developed as part of the MAGUS AI ecosystem.
Our goal is to build proprietary technology capable of understanding, reasoning, and assisting millions of people and organizations, reducing dependence on external models and driving technological innovation from our region. This project represents a long-term investment in infrastructure, research, model training, and specialized talent.
The following timeline shows the main stages of research and development we will follow until the launch of the first generation of the AI model created by MAGUS project BigBox, an initiative that aspires to become one of the most important private artificial intelligence projects in Latin America.
Definition of the project vision, the objectives of the AI model, and the technologies that will serve as the foundation for its development.
Design of the internal structure of the AI model, including its architecture, capabilities, and training strategy.
Implementation of the high-performance infrastructure required to train large-scale artificial intelligence models.
Configuration of frameworks, libraries, tools, and systems that will enable the training and development of the model.
Collection, cleaning, and organization of large-scale datasets to ensure high-quality training.
Initial training using deep learning techniques to develop the model’s core capabilities.
Application of fine-tuning, alignment, and optimization to improve the quality, accuracy, and usefulness of the responses.
Evaluation using benchmarks and tests of quality, reasoning, safety, and efficiency prior to release.
A limited group of users tests the model to detect errors, validate features, and gather feedback.
The model is made available to a broader group of users to continue its optimization through real-world use cases.
Release of the first official version of MAGUS AI Suite's proprietary model, making it available to all users.
New versions, continuous training, and ongoing improvements to expand the capabilities of the MAGUS Model over time.
Meet MAGUS
Be part of the future of MAGUS
Our research project continues to grow. If you are a researcher, engineer, or simply share our vision of building artificial intelligence from Latin America, we would love to hear from you.
Contact us
[email protected]
Español