TLDR: SuperX AI Technology Limited has launched its new All-in-One Multi-Model Server (MMS) series, designed to provide enterprises with out-of-the-box, secure, and efficient full-stack AI solutions. These servers come pre-configured with OpenAI’s latest open-source large language models, GPT-OSS-120B and GPT-OSS-20B, aiming to redefine enterprise AI productivity through multi-model collaboration and rapid deployment.
SINGAPORE – SuperX AI Technology Limited (NASDAQ: SUPX) today announced the official launch of its groundbreaking All-in-One Multi-Model Server (MMS) series, marking a significant advancement in enterprise AI infrastructure. This new series is engineered to deliver immediate, comprehensive AI capabilities to businesses of all sizes, from individual professionals to large enterprises.
The MMS series is distinguished by its ‘out-of-the-box’ readiness, featuring pre-configured integration with OpenAI’s newly released, high-performance large language models (LLMs), GPT-OSS-120B and GPT-OSS-20B. These models are critical for enabling world-class AI inference and knowledge-processing capabilities at superior cost efficiency. Benchmark results indicate that GPT-OSS-120B achieves, and in some key tests like Massive Multitask Language Understanding (MMLU) and American Invitational Mathematics Examination (AIME), even surpasses the performance of several leading closed-source models, according to OpenAI’s press release on August 5, 2025.
SuperX’s MMS is designed for multi-model fusion and deep integration into various application scenarios, supporting the dynamic collaboration of diverse AI models including inference, general-purpose, speech synthesis/recognition, and text-to-image models. This comprehensive platform eliminates the need for lengthy model integration and hardware adaptation, offering a streamlined solution for deploying generative AI applications.
Security is a cornerstone of the new MMS series, which incorporates advanced data security protocols through NVIDIA Confidential Computing. This technology provides a Trusted Execution Environment (TEE) to protect AI intellectual property and secure AI model training and inference processes. Additionally, the series supports over 60 pre-configured scenario-based agents and facilitates rapid deployment without requiring extensive IT resources.
“A single model cannot solve the problems of a complex world. Multi-model collaboration is a vital step in the evolution of AI,” stated Kenny Sng, CTO of SuperX, emphasizing the strategic importance of the new server series.
The MMS product line offers customized specifications to cater to different enterprise scales, with pricing options ranging from $50,000 for the AI Workstation Standard to $4,000,000 for the Cluster Edition. Specifically, cluster editions start at $4 million for large enterprises, a B200 Standard Edition is priced at $500,000 for medium-sized businesses, and AI workstation versions are available from $50,000 to $250,000 for individual professional use.
Also Read:
- OpenAI and NVIDIA Unveil New Open-Weight AI Models for Global Inference Infrastructure
- OpenAI Unveils New Open-Weight Models, Expanding Accessibility and Cloud Partnerships
This launch follows SuperX’s recent introduction of its XN9160-B200 AI Server on July 30, 2025, which is powered by NVIDIA Blackwell B200 GPUs. The XN9160-B200 is engineered to address the computational demands of generative AI, LLMs, and high-performance computing (HPC), delivering inter-GPU bandwidth of 1.8TB/s—30 times faster than the previous H100 platform. The company’s strategic focus on AI infrastructure is further underscored by its rebranding from Junee Limited and the establishment of a new Japan supply center, aiming to produce 10,000 units annually by late 2025 to meet surging demand in Asia-Pacific markets. SuperX is targeting high-value enterprise segments such as finance, healthcare, and government sectors with these advanced AI solutions.
