Malaysia has taken a bold step into the AI spotlight with the launch of Ilmu 1.0, the nation’s first homegrown multimodal AI model capable of processing and generating text, voice, and images — all fine-tuned for Malay, English, and local dialects.
Unveiled by Prime Minister Datuk Seri Anwar Ibrahim at the Asean AI Summit 2025, Ilmu — short for Intelek Luhur Malaysia Untukmu — is the result of a collaboration between YTL AI Labs and Universiti Malaya.
According to Anwar, the project aligns with Malaysia’s ambition not just to keep pace with global AI advancements, but to lead in areas that reflect the country’s values of equity and justice. He also emphasised the importance of making such technology accessible to rural and marginalised communities.
Multimodal AI models are designed to work across different types of data simultaneously — from generating captions for images to answering questions about videos — making them more capable of handling real-world tasks. YTL claims Ilmu 1.0 performs on par with heavyweights like GPT-4o and Llama 3.1, while excelling in Malay language comprehension and locally relevant problem-solving.
The model is hosted on YTL’s sovereign AI cloud for enterprise-grade security and data residency, ensuring sensitive information stays within national borders. To fuel adoption, YTL AI Labs has launched the Ilmu AI Accelerator Programme with the Malaysian Digital Economy Corporation, offering RM5 million to startups and developers building tools powered by Ilmu. Consumer access through ILMUchat will roll out on Malaysia Day, 16 September.
YTL has committed over RM20 billion towards Malaysia’s AI ecosystem, covering green data centres, sovereign models, and partnerships with Astro, Media Prima, Ryt Bank, Carsome, and Swipey. Digital minister Gobind Singh Deo called Ilmu a groundbreaking initiative that anchors Malaysia’s push for ethical, locally relevant AI development.
