Skip to content

SouthwardTech

Technology, industry and investment across China and the Global South

5 projects tracked1 Contracted4 AnnouncedReviewed

Global South

China’s Z.ai lends its model to Malaysia’s sovereign AI push

The Chinese lab's GLM-5.3 underpins YTL's new coding platform, priced slightly below Z.ai's own rates and, YTL says, hosted in Malaysia

China’s Z.ai, also known as Zhipu AI, has become a partner in Malaysia’s sovereign AI push. YTL AI Labs launched ILMUcode at Universiti Malaya on 1 October, powered by ILMU-GLM-5.3, a model it says was built in partnership with Z.ai.

Four days later, Amazon Web Services made GLM 5.3 available to eligible enterprise customers on its Bedrock platform, served through cross-Region inference, which routes requests among several AWS regions. YTL’s model documentation instead says the models in its API catalogue, which lists both GLM services, are “hosted on Malaysian infrastructure”.

YTL said 800 first-year computer science students at the university would each receive RM100 in monthly credits for three months. The service is also open to the public through ilmucode.ai.

ILMU’s listed API prices for the model are lower than Z.ai’s own. It charges RM5.15 per million input tokens and RM16.05 per million output tokens, before tax. At Bank Negara Malaysia’s 16 October, 5pm middle rate of RM4.082 to the US dollar, that is about US$1.26 and US$3.93, roughly 10 per cent below Z.ai’s US$1.40 and US$4.40. Adding ILMU’s 8 per cent sales and service tax narrows the gap to roughly 3 per cent, before any tax a Z.ai customer might owe.

YTL’s own models remain part of the offering. Its price list includes ILMU v3.1 and ILMU Mini v3.3 alongside GLM; Universiti Malaya professor Chan Chee Seng described the original ILMU in September 2025 as built from scratch. In the launch statement, chief executive Foong Chee Mun said the company was building its flagship models while “partnering with the world’s leading AI companies from NVIDIA to Z.ai”.

Z.ai published GLM-5.3’s weights in late August. YTL’s launch statement does not specify how ILMU-GLM-5.3 relates to that release, whether YTL further trained the model or how technical work was divided between the partners. By contrast, AI Singapore identifies Qwen as the base model in both the name and model card of its SEA-LION variant.

Higher Education Minister Zambry Abdul Kadir said ILMUcode would be extended to other higher education institutions, Bernama reported; no timeline was given. YTL gave no active-user or paying-customer figures in its launch announcement.

Published 17 October 2026 · Last updated 17 October 2026 · Corrections to corrections@southwardtech.com

Weekly newsletter

The Southward Brief

A weekly briefing on the technologies, companies and industrial shifts reshaping China and the Global South.

Required. Enter a valid email address.

By subscribing, you agree to receive The Southward Brief. You can unsubscribe at any time. Read our Privacy Policy.