One platform connecting startups, corporations, research institutions, regions and investors
Playbook & Handbook · February 6, 2026
Winning Accelerator Playbook 2026 by BambuUP
EDTECH: LEVERAGING AI-POWERED TECHNOLOGIES TO TRANSFORM THE LEARNING EXPERIENCE
1. Technology overview
ILMU is a family of AI models and products from YTL AI Labs Sdn Bhd, the entity named in the copyright line of ilmu.ai. The offering centres on a Malaysian-built language model family accessed through chat apps and an API console.
The model catalogue lists the flagship ILMU model (API string ilmu-v3.1), ILMU-Nemo-Nano for agentic tasks, ILMU Vision, ILMU ASR for speech-to-text, ILMU TTS for text-to-speech, plus embedding, reranker and web search models. Products include ILMUchat, ILMUcode and ILMUclaw agents. Partners listed include NVIDIA, Ryt Bank, MDEC, Universiti Malaya and Razorpay Curlec.
2. Highlighted features & specifications
- ILMU LLM: flagship language model available by API as ilmu-v3.1.
- ILMU ASR and TTS: speech-to-text and text-to-speech models in the same catalogue.
- ILMU Vision: image understanding model.
- Embedding and reranking: ILMU Embedding, BGE-M3 and BGE Reranker for retrieval.
- ILMUcode: AI coding platform launched on 1 October 2026.
- ILMUclaw: autonomous agents on OpenClaw-compatible infrastructure.
3. Potential application
Malay-language customer-service assistants for Vietnamese firms selling in Malaysia. Speech-to-text and TTS for Malay call-centre automation. Retrieval-augmented search over Malay documents using the embedding and reranker models.
4. Benefits
A single vendor supplies LLM, speech recognition, speech synthesis, vision and embedding models through one API console. Models are built in Malaysia for local language needs, with partners in banking and government digital programmes. Each model is exposed through a stable API model string for integration.
5. Market trends & opportunities
Southeast Asian governments and corporates are backing sovereign language models for Bahasa Malaysia and regional dialects. Banks and public bodies value local models for data-residency and language coverage, and NVIDIA is supporting regional model and dataset releases.
6. Other information
Operated by YTL AI Labs Sdn Bhd (per site footer). News on the site: ILMUcode launch on 2026-10-01 and the Nemotron-Personas-Malaysia dataset with NVIDIA on 2026-09-23. Partners include NVIDIA, Ryt Bank, Geneco, Ranhill, MRANTI, MDEC and Universiti Malaya.
