ILMU (Malaysian Large Language Model)

ILMU is described as Malaysia's first homegrown large language model, developed by YTL AI Labs, a research initiative of the YTL group. Announced in August 2025 with early access to a chat application beginning around Malaysia Day in September 2025, ILMU is a multimodal model built to understand and respond in Bahasa Melayu, Manglish and regional dialects across text, voice and vision. Its development positions it as part of Malaysia's effort to build sovereign artificial intelligence aligned with local language, culture and data-residency requirements.

Name and design philosophy

The name ILMU, which itself means knowledge in Malay, is presented as an acronym reflecting the model's stated values: Intelek for context-aware intelligence, Luhur for an ethical foundation rooted in Malaysian values, Malaysia for cultural and linguistic fluency, and Untukmu, meaning for you, signalling that it is built to serve Malaysians. The framing emphasises local relevance over a generic global model, with attention to the languages, idioms and cultural context of Malaysian users.

Capabilities

ILMU is a multimodal system, meaning it processes and produces more than text. According to its developers it handles text, voice and vision, and it is designed to understand Bahasa Melayu together with the code-switching of Manglish and regional dialects such as Kelantanese. The model was developed in collaboration with Universiti Malaya, drawing on local academic expertise. Reported benchmark results place ILMU as a strong performer on Malay-language evaluations such as the Malay MMLU, where its developers say it outperforms larger international models on Malay and dialect tasks, although independent verification of such claims is part of the normal scrutiny applied to any new model.

Sovereignty and infrastructure

A central theme of ILMU is data sovereignty. Its developers state that data is stored within Malaysia to protect user privacy and maintain national control over the AI stack. This is supported by YTL's broader investment in domestic infrastructure, including local data centres and AI-cloud capacity built in partnership with Nvidia. By owning the model, the data and the compute, the initiative aims to reduce reliance on foreign AI services for sensitive Malaysian applications, a goal consistent with national policy on digital self-determination.

| Aspect | Detail | | --- | --- | | Languages | Bahasa Melayu, Manglish, dialects | | Modalities | Text, voice, vision | | Data residency | Stored within Malaysia | | Infrastructure | Local data centres, AI cloud with Nvidia |

ILMU sits at the centre of Malaysia's sovereign-AI agenda, complementing other local efforts such as MaLLaM and aligning with the goals of the MyDigital Blueprint and the work of the National AI Office. By building a model that is fluent in Bahasa Melayu and regional dialects and that keeps data within the country, YTL AI Labs addresses a gap left by international models trained predominantly on English and global data.

The initiative reflects a wider national strategy in which agencies including MDEC and the Ministry of Digital promote domestic AI capability, while data-residency concerns echo the principles of the Personal Data Protection Act and emerging AI governance guidance. YTL's investment in data centres and AI-cloud infrastructure, including facilities developed with Nvidia in Johor, gives Malaysia the compute base needed to train and serve large models locally rather than depending on overseas providers.

For Malaysian enterprises in banking, government and telecommunications, a sovereign model fluent in local languages offers a path to deploying generative AI for customer service, document processing and analytics while keeping sensitive information onshore. The project also strengthens the local AI talent ecosystem through its collaboration with Universiti Malaya and contributes to the skills agenda supported by HRD Corp, helping Malaysia build expertise that extends across the broader Southeast Asian market.

1. YTL AI Labs. (2025). ILMU: Malaysia's AI — Multimodal, Sovereign, Built for Malaysians. ytlailabs.com.
2. The Edge Malaysia. (2025). YTL Power launches Malaysia's first homegrown LLM, ILMU. theedgemalaysia.com.
3. Malay Mail. (2025). Meet ILMU, Malaysia's first homegrown AI model. malaymail.com.