MBZUAI Launches Institute Of Foundation Models With New Silicon Valley AI Lab For Enhanced Research Collaboration

Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) has introduced the Institute of Foundation Models (IFM). This initiative spans multiple locations, including a new lab in Sunnyvale, California, and existing facilities in Paris and Abu Dhabi. The launch at the Computer History Museum in Mountain View marks the third node in MBZUAI's global research network, linking it with California's dynamic AI community.

The UAE's strategic move to invest in advanced AI foundation models aligns with its long-term economic diversification goals. By fostering knowledge-based sectors, the country aims to support its economic and social transformation. "The launch of the IFM represents a major step forward for the collaboration and global development of frontier-class AI foundation models," said Professor Eric Xing, President and University Professor of MBZUAI.

The expansion into Silicon Valley enhances MBZUAI's presence in one of the world's most active AI ecosystems. This strategic location facilitates knowledge exchange with leading institutions and access to a talent pool skilled in scaling research into practical applications. The launch event attracted representatives from top AI companies and academic institutions, underscoring interest in MBZUAI's global approach to foundation model research.

At the core of MBZUAI's demonstrations was PAN, a world model capable of infinite simulations across diverse realities. Unlike previous systems that focused on generating text, audio, or images, PAN predicts comprehensive world states by integrating multimodal inputs like language, video, spatial data, and physical actions. This enables advanced reasoning and decision-making for applications such as autonomous driving and robotics.

PAN’s hierarchical architecture supports multi-level reasoning and real-time interaction within simulations while maintaining high accuracy over extended scenarios. Its companion system, PAN-Agent, demonstrates its utility in multimodal reasoning tasks like mathematics and coding within dynamic simulated environments.

The IFM lab is also advancing two flagship AI systems: K2 and JAIS. An upcoming update to K2-65B will focus on delivering breakthrough reasoning capabilities with sustainable performance. K2 aims to enhance mathematical problem-solving, code generation, and logical analysis while using fewer computational resources than many comparable models.

Promoting Language Diversity

JAIS is recognised as the world's most advanced Arabic large language model. This open-sourced system addresses the underrepresentation of non-English languages in AI by covering Modern Standard Arabic, regional dialects, Hindi, and other languages while maintaining cultural authenticity. At IFM, JAIS will continue to expand its capabilities with increased language support.

MBZUAI has established one of the industry's most transparent approaches to AI development by open-sourcing not just models but entire development processes—positioning IFM as a leader in building openly. The LLM360 initiative provides researchers with complete materials including training code, datasets, and model checkpoints.

Commitment to Open Development

This openness is balanced with safeguards including international advisory boards and peer review processes that maintain research integrity. The IFM's structure includes dedicated teams focused on model architecture, training methods, evaluation frameworks, and safety systems—combining startup agility with established research institution resources.

Through partnerships with industry leaders, academic institutions, and public organisations, IFM is building a framework to translate research into practical applications globally advancing the field of AI.

With inputs from WAM