Sintérgica AI
Sintérgica Labs · Open Research

Sintérgica Labs: Frontier research from Mexico

Non-profit research lab. We position Mexico and Latin America at the frontier of AI knowledge.

Collaborate with LabsView research lines
Non-profitOpen researchMexico & LATAMOpen collaboration
released models
4
papers & research
10+
academic partners
5+
Open science

Collectively advancing the science of AI in LATAM

01

Open Science Initiative

By building an open science community and releasing our models, we are empowering the next era of technological progress in the region.

View on HuggingFace
02

Data Sovereignty

Curation of high-quality datasets with regulatory, cultural, and linguistic context from Mexico and Latin America to train models without WEIRD bias.

03

Institutional Collaboration

We provide access to infrastructure and models to academic partners and civic institutions for projects driving real-world impact through AI.

Anchor project

Lattice Na’at: toward the global frontier

Sintérgica AI developed the largest AI model in LATAM with 1 trillion parameters. This is the path to the global frontier.

Lattice Na’at1T Parameters · Active
New Model

Lattice Na'at 1T

Largest foundational model built in LATAM. 1 trillion (1T) parameters trained on a regulatory, cultural, and linguistic corpus from Mexico and Latin America.

View full modelView on HuggingFace
  1. Active2025

    1T

    Lattice Na’at

    Largest AI model developed in LATAM. Regulatory, cultural, and linguistic knowledge base for Mexico and LATAM. Active production.

  2. In development2027

    1T

    Na’at Scale

    Advanced MoE architecture

    Scaling toward 1 trillion parameters with advanced MoE architecture. Enhanced reasoning and extended context.

  3. Goal2030

    Frontier

    Na’at Global

    Position a Latin American-origin model at the global frontier of AI knowledge. Regional cognitive sovereignty.

Sintérgica AI implements private AI for regulated sectors in Mexico and Latin America.

Explore Na’at in detail
Research lines

Nine frontier research lines

Applied research focused on real AI challenges for non-Anglo-Saxon contexts.

Efficient modelsData & languagesEvaluationReasoning
  1. 01Efficient models

    Efficient knowledge distillation

    Transfer of capabilities from large models (1T+) to compact models without significant performance loss.

  2. 02Efficient models

    Low-cost fine-tuning (LoRA, QLoRA)

    Parameter-efficient fine-tuning methods for adapting base models to specific domains with limited resources.

  3. 03Data & languages

    Dataset curation for MX and LATAM

    Building high-quality corpora with regulatory, cultural, and linguistic context from Mexico and Latin America.

  4. 04Data & languages

    NLP for indigenous languages

    Models and tools for natural language processing of Nahuatl, Maya, and other indigenous languages of Mexico.

  5. 05Evaluation

    Non-WEIRD benchmarks

    Spanish HELM and MMLU-LatAm: model evaluation with metrics that don't assume Western, Educated, Industrialized context.

  6. 06Evaluation

    Cultural bias mitigation

    Identification and reduction of WEIRD biases in models trained predominantly on Anglo-Saxon data.

  7. 07Reasoning

    Tensorial logical representations

    Encoding of structured knowledge and logical relations in tensorial spaces for explicit reasoning.

  8. 08Reasoning

    Efficient recursive reasoning

    Architectures and strategies for multi-step reasoning with computational efficiency at inference time.

  9. 09Reasoning

    SLMs for agentic systems

    Specialized small models as components of multi-agent systems with verifiable governance.

Our Research

Publications and Models

View all publications
Paper

WEIRD bias mitigation in LLMs

Coming soon

Evaluation

MMLU-LatAm: Evaluating regional context

Coming soon

Model

Lattice Séeb 9B

2025

Collaborate

Join the research

Sintérgica Labs is an open project. We seek collaborators who want to build the future of AI from Latin America.

Researchers

Access & publication

  • Early access to models and datasets
  • Co-authorship in publications
  • Shared compute infrastructure
  • LATAM researcher network

Universities

Partnerships & projects

  • Formal collaboration agreements
  • Thesis and graduate projects
  • Joint workshops and seminars
  • Research grants

Developers

Open source & APIs

  • Contribute to open source models
  • Research APIs (early access)
  • Bounties for technical improvements
  • Technical mentorship

Organizations

Funding & data

  • Contribution of specialized data
  • Funding of research lines
  • Real use cases as benchmarks
  • Recognition in publications
Contact the Labs team