LANTA operates on a cutting-edge Heterogeneous HPE Cray EX system, delivering a peak performance of 8.15 petaflops. The robust architecture comprises 346 compute nodes, 31,744 CPU cores, and 704 NVIDIA A100 GPUs. This immense computational power is supported by a 200 Gbps HPE Slingshot network and a 10-petabyte parallel storage system, making it highly optimized for AI training, massive data analytics, and complex parallel processing.
The performance leap is substantial. Training an eight-billion-parameter Large Language Model from scratch would traditionally take over 160 years on a single GPU or two decades on an eight-GPU workstation. Utilizing LANTA's parallel processing capabilities with approximately 700 GPUs reduces this timeframe to merely 87 days.
Beyond raw computing power, LANTA represents a strong commitment to sustainable technology. The system utilizes advanced liquid cooling technology, which reduces cooling energy consumption by 33 percent compared to conventional air-cooled systems, marking a first for Thailand. This eco-friendly approach earned LANTA the 24th global position on the Green500 list in 2022 and the 91st position in 2025. It also achieved significant global recognition on the TOP500 list, ranking 70th globally in 2022 and 199th in 2025, while holding the top position in ASEAN for both performance and sustainability during its debut year.
The infrastructure has been the driving force behind numerous high-impact national and international projects. In the realm of artificial intelligence, LANTA powered the development of Pathumma LLM, an open-source generative AI, and ThaiLLM, the national foundation model tailored to the Thai language and culture launched earlier this year. In healthcare, the Medical AI Consortium utilizes the system to build a centralized diagnostic AI platform, while the DMIND depression screening application relies on models trained on LANTA to serve the public through the national health application.
Environmental and agricultural initiatives also benefit significantly from this computational power. The Pollution Control Department revolutionized its PM2.5 forecasting capabilities, shifting from a three-day forecast requiring 22 hours of processing to a highly accurate seven-day forecast generated in just 45 minutes, enabling timely public health alerts. Additionally, the Asian Institute of Technology utilizes the system for ultra-high-resolution, six-month weather forecasting to advance precision agriculture. The center also plays a pivotal role in human resource development, supporting the Super AI Engineer initiative which trains over 1,000 specialists annually, alongside providing comprehensive HPC training and system architecture consulting for global organizations.
The National Supercomputer Centre offers scalable services tailored to public and private legal entities. The flexible pricing model is based on Service Node-Hours, with project access starting at an accessible rate of 5,000 THB. The service includes comprehensive computational software and expert technical guidance. To foster continuous innovation, the center provides dedicated support for Proof of Concept testing and periodically offers research voucher programs. Organizations seeking to leverage this advanced computing infrastructure can access detailed information and service inquiries through the official ThaiSC website at www.thaisc.io and the ThaiSC Facebook page.