About the National Multimodal LLM Programme
In partnership with AI Singapore (AISG) and the Agency for Science, Technology and Research (A*STAR), IMDA launched Singapore’s National Multimodal Large Language Model (LLM) Programme (NMLP) in 2023 to drive regional and domestic innovation in AI, build the nation’s capabilities in AI research & development (R&D), and develop AI talent. This programme marks a significant leap forward in building the next bounds of AI capabilities for Singapore and the region.
Funded by the National Research Foundation, Singapore (NRF), the S$70 million Initiative will develop Singapore’s research and engineering capabilities in multi-modal LLMs. In line with the Research, Innovation and Enterprise (RIE) 2025 plan, the two-year national-level initiative will also support Singapore’s National AI Strategy 2.0 (NAIS 2.0).
The NAIS 2.0 guides Singapore to become a global leader in AI solutions by 2030 through the development and deployment of scalable, impactful AI solutions in key sectors that are high value and relevant to citizens and businesses.
The NMLP will:
- Build skilled AI talent in Singapore with funding support and access to high-end computing for local researchers and engineers, enabling the growth of a talent pool that drives innovation and technological excellence in AI development.
- Foster a thriving AI industry to develop LLM-enabled solutions that boost productivity and unlock new opportunities for businesses, while promoting industry collaboration to create a conducive environment for innovative enterprise use cases in a rapidly evolving AI landscape.
- Establish a trusted environment for AI use by deepening understanding of how LLMs work and advancing research in AI governance. The NMLP will also develop a base model grounded in regional context, capable of understanding the unique linguistic and multilingual characteristics of Singapore and the region.
Models Developed
MERaLiON
Developed by A*STAR Institute for Infocomm Research (A*STAR I2R), MERaLiON recognises conversations that blend languages and dialects commonly used across cultures in Singapore and Southeast Asia.
Through its multimodal integration, MERaLiON advances the understanding of human communication dynamics – propelling AI capabilities significantly.
The first phase of MERaLiON leverages multimodal and multilingual representation learning and alignment to enable more effective training and improve model generalisation, allowing it to comprehend colloquial language and perform downstream tasks more accurately.
Its features include:
- Multilingual Speech Transcription and Translation
- Speech Summarisation
- Speech Question and Answer
- Audio Scene Understanding
- Paralingual Understanding
- Local Speech Understanding
MERaLiON can empower public agencies and businesses to address current challenges and encourage the development of innovative use cases that drive future growth and efficiency, particularly in the areas of customer service experiences, discovery of new insights, and agentic decision making.
SEA-LION
Developed by AI Singapore, SEA-LION models are LLMs specifically trained on content in Southeast Asian languages and dialects, including Thai, Vietnamese, and Bahasa Indonesia. By capturing the region’s linguistic diversity and cultural nuances, the SEA-LION models deliver more accurate and contextually relevant responses than those primarily trained on Western or Chinese data.
SEA-LION models are designed to address the underrepresentation gap, advance inclusive AI, and better serve the needs of Southeast Asia’s diverse populations.
Current SEA-LION models are already functioning for the Southeast Asia region, outperforming other open-source models of similar size. For more information, please visit https://sea-lion.ai/.
How you can collaborate
Collaborating with A*STAR I2R (MERaLiON)
Join the MERaLiON Ecosystem and tap into the collective intelligence of the MERaLiON ecosystem with rich data sources, use cases, and expertise. Partner with us to unlock powerful network effects that drive exponential growth.
Deepen your involvement and shape the future of multimodal AI in Southeast Asia by engaging with the MERaLiON AI Consortium. Spearheaded by A*STAR’s Institute for Infocomm Research (A*STAR I²R) and supported by IMDA, the Consortium serves as a key platform where ecosystem knowledge and expertise are harnessed. Together, we build on the MERaLiON model and co-develop it into impactful use cases, accelerating adoption. The first group of members, including Axiom IT Solutions, BytePlus, CommonTown, DBS Bank, Grab, HTX, Microsoft, MOH Office for Healthcare Transformation (MOHT), NCS, National Supercomputing Centre (NSCC) Singapore, SPH Media and ST Engineering, will be exploring diverse use cases such as multilingual customer support, health & emotional insight detection, and agentic decision-making. This consortium approach helps to accelerate early adoption by:
- Aggregating demand to reduce development and integration costs.
- Pooling resources (such as data, expertise) to improve model capabilities.
- Providing training and tools to build enterprise readiness.
- Co-developing real-world use cases and early Proofs of Concept (POCs) to demonstrate Return on Investment (ROI).
To learn more, visit www.meralion.ai or reach out to nmlp@imda.gov.sg.
Collaborating with AI Singapore (SEA-LION)
AI Singapore welcomes collaborations to further enhance and expand the capabilities of SEA-LION. Offering various partnership modalities to suit different needs and interests, AI Singapore focuses on open-sourcing results to benefit the wider community.
To explore opportunities, reach out to nmlp@imda.gov.sg.
.webp)