One-Sentence Answer
ILMU Console is Malaysia's answer to data sovereignty in the AI era — a homegrown inference platform that keeps your data local, understands your multilingual context, and integrates seamlessly with existing OpenAI-compatible code.
The Problem: AI APIs That Don't Understand Malaysia
When Malaysian developers build AI-powered applications, they face three recurring frustrations:
-
Data leaves the country. Every API call to OpenAI, Anthropic, or Google sends data to US or EU servers. For healthcare, finance, or government projects, this violates PDPA compliance and data sovereignty requirements.
-
The AI doesn't understand us. Ask ChatGPT about "kopi O kosong" or "Bukit Bintang footfall" and you get confused responses. Western models weren't trained on Malaysian English, Bahasa Malaysia, or our local business context.
-
Pricing is in USD. At RM 4.70 per USD, a $20/month OpenAI subscription costs RM 94. Currency fluctuations make budgeting unpredictable, especially for long-term projects.
ILMU Console, built by YTL AI Labs, solves all three problems with a single platform.
What Is ILMU Console?
ILMU Console is a sovereign AI inference platform hosted entirely on Malaysian infrastructure (YTL AI Cloud). It provides:
- OpenAI-compatible APIs — use your existing SDK code, just change the endpoint
- 100% data residency — your data never leaves Malaysia
- Local language understanding — trained on Malaysian English, Bahasa Malaysia, and Chinese with local nuance
- Ringgit-based pricing — no foreign exchange risk
- Multiple model options — from lightweight Nemo-Nano to high-performance Nemo-Super
The platform launched in 2025 and has grown to over 5,400 developers building agents, chatbots, and enterprise applications on Malaysian soil.
Key Features for Malaysian Developers
1. Multi-SDK Compatibility
ILMU exposes three API endpoints that accept the same sk- prefixed API key:
| SDK Format | Base URL |
|------------|----------|
| OpenAI | https://api.ilmu.ai/v1 |
| Anthropic | https://api.ilmu.ai/anthropic |
| Gemini | https://api.ilmu.ai/gemini |
This means you can migrate existing projects by changing one line of configuration:
// Before (OpenAI)
const openai = new OpenAI({
apiKey: process.env.OPENAI_API_KEY,
});
// After (ILMU)
const openai = new OpenAI({
apiKey: process.env.ILMU_API_KEY,
baseURL: 'https://api.ilmu.ai/v1',
});
No SDK changes, no wrapper libraries, no vendor lock-in.
2. Malaysian Context Understanding
ILMU's models are trained on local data and understand:
- Manglish and Bahasa Rojak — "Can you help me check the SSM registration for this company?"
- Local business context — kopitiam operations, SME supply chains, government procurement processes
- Cultural nuance — knows that "Georgetown" refers to Penang, not the US city
- Multilingual support — seamless switching between English, Malay, and Chinese in the same conversation
This is particularly valuable for customer-facing applications where Western models produce awkward or incorrect responses.
3. Sovereign Data Residency
For regulated industries, data sovereignty is non-negotiable. ILMU guarantees:
- All inference runs on YTL AI Cloud servers in Malaysia
- No data is used for model training (unless you explicitly opt in)
- Compliant with PDPA (Personal Data Protection Act 2010)
- Suitable for government, healthcare, finance, and education sectors
This makes ILMU the only viable option for projects that cannot send data overseas.
4. Agent-Ready Infrastructure
ILMU Claw plans are built for autonomous AI agents, not just chatbots:
- Tool calling and function calling — JSON tool responses compatible with OpenAI format
- Streaming responses — SSE (Server-Sent Events) for real-time token-by-token output
- High-performance models — Nemo-Super ranks #1 on PinchBench (85.6% score) for agent tasks
- Guaranteed allocations — no peak-hour throttling, consistent performance
If you're building with frameworks like LangChain, CrewAI, or AutoGen, ILMU works out of the box.
Pricing: Transparent and Ringgit-Based
ILMU offers two product lines: Claw (for agents and general inference) and Coding (for IDE integrations).
ILMU Claw Plans
| Plan | Price | Models | Best For | |------|-------|--------|----------| | Free | RM 0 | Nemo-Nano, Nemo-Super | Testing and prototyping | | Starter | RM 25/month | Nemo-Nano, Nemo-Super, BGE-M3 | Small projects and MVPs | | Pro | RM 50/month | All Claw models | Production workloads |
All plans include:
- Streaming responses
- Tool calling
- 100% data residency
- No training on your data
Top-up packs are available for burst usage:
- 35M tokens: RM 30 (one-time)
- 100M tokens: RM 60 (one-time)
Tokens from top-ups last 12 months, giving you flexibility without monthly commitment.
Pay-As-You-Go (PAYG)
For projects that need models beyond Claw (vision, embeddings, speech):
| Model | Type | Input Rate | Output Rate | |-------|------|------------|-------------| | ILMU v3.1 | Text | RM 4.00/1M tokens | RM 16.00/1M tokens | | ILMU Vision v1.3 | Vision | RM 1.60/1M tokens | RM 4.80/1M tokens | | ILMU Mini v3.3 | Text (lightweight) | RM 0.20/1M tokens | RM 1.20/1M tokens | | BGE-M3 | Embeddings | RM 0.04/1M tokens | — | | ILMU ASR v4.2 | Speech-to-text | RM 0.0002/sec | — | | ILMU TTS v2 | Text-to-speech | RM 0.08/1K chars | — |
PAYG credit is shared across your organization and valid for 12 months.
Note: Effective June 1, 2026, all paid plans are subject to 8% SST (Sales & Service Tax).
Who Should Use ILMU Console?
Malaysian Businesses with Data Sovereignty Requirements
If your project involves:
- Government contracts (MAMPU guidelines)
- Healthcare data (patient records, medical imaging)
- Financial services (banking, insurance, fintech)
- Education (student data, exam systems)
ILMU is the only platform that guarantees Malaysian data residency.
SMEs Building AI-Powered Products
For Malaysian SMEs building:
- Customer service chatbots that understand Manglish
- Document processing for local forms and contracts
- Internal tools for multilingual teams
- E-commerce recommendation engines
ILMU's local context understanding and Ringgit pricing make it more cost-effective than foreign alternatives.
Developers Building AI Agents
If you're building autonomous agents with:
- LangChain, CrewAI, AutoGen, or similar frameworks
- Tool calling and function calling
- Multi-step reasoning workflows
- Real-time streaming responses
ILMU Claw plans are purpose-built for agent workloads with guaranteed performance.
Getting Started (5 Minutes)
- Sign up at console.ilmu.ai
- Generate an API key from the Dashboard
- Configure your SDK to use ILMU's base URL
- Make your first API call
Example using the OpenAI SDK:
import OpenAI from 'openai';
const client = new OpenAI({
apiKey: process.env.ILMU_API_KEY,
baseURL: 'https://api.ilmu.ai/v1',
});
const response = await client.chat.completions.create({
model: 'nemo-super',
messages: [
{
role: 'user',
content: 'Analisis pasaran: Kopitiam premium di Bukit Bintang vs Bangsar.',
},
],
stream: true,
});
for await (const chunk of response) {
process.stdout.write(chunk.choices[0]?.delta?.content || '');
}
The response will demonstrate ILMU's understanding of Malaysian business context — comparing footfall patterns, demographic differences, and local market dynamics that a Western model would miss.
Comparison: ILMU vs OpenAI vs Anthropic
| Feature | ILMU Console | OpenAI | Anthropic | |---------|--------------|--------|-----------| | Data residency | 100% Malaysia | US/EU | US/EU | | Local language support | Excellent (Manglish, BM, Chinese) | Good (English) | Good (English) | | Pricing currency | MYR | USD | USD | | Agent optimization | Yes (Nemo-Super, 85.6% PinchBench) | Yes (GPT-4) | Yes (Claude 3.5) | | OpenAI SDK compatible | Yes | Yes | No | | Malaysian context | Deep | Limited | Limited | | Peak-hour throttling | None (guaranteed allocation) | Possible | Possible |
For Malaysian projects, ILMU wins on sovereignty, local understanding, and pricing predictability. For global projects with no data residency requirements, OpenAI and Anthropic remain strong choices.
Limitations to Consider
ILMU is excellent for Malaysian use cases, but has some constraints:
- Smaller model ecosystem — fewer model options than OpenAI or Anthropic (though the available models are well-optimized)
- Regional latency — optimal for Southeast Asia; users in US/EU may experience higher latency
- Limited free tier — 200 seats available, requires phone verification
- Newer platform — less community documentation and third-party integrations compared to OpenAI
These limitations are acceptable trade-offs for the sovereignty and local context benefits.
Conclusion
ILMU Console represents a significant milestone for Malaysia's AI ecosystem. For the first time, Malaysian developers have a sovereign alternative to foreign AI platforms — one that understands our languages, respects our data sovereignty laws, and prices in our currency.
For businesses in regulated industries (government, healthcare, finance), ILMU is not just an option — it's a requirement. For SMEs and startups, it's a cost-effective way to build AI-powered products with local context.
The platform's OpenAI-compatible API design means migration is trivial: change one line of configuration and your existing code works. The multi-SDK support (OpenAI, Anthropic, Gemini formats) gives you flexibility without vendor lock-in.
As Malaysia's AI adoption accelerates, platforms like ILMU Console will become critical infrastructure. The question is not whether you'll need sovereign AI — it's whether you'll adopt it early enough to build competitive advantage.
Extended Reading
- ILMU Console official website
- ILMU API documentation
- YTL AI Labs — the team behind ILMU
- Hermes Agent — an open-source AI agent that can integrate with ILMU as a provider
