SAN FRANCISCO, July 23, 2026 /PRNewswire/ — Moreh, an AI infrastructure software company, led by CEO Gangwon Jo, participated in AMD Advancing AI 2026, AMD’s flagship annual AI event held in San Francisco on July 22–23 (local time), where it demonstrated its distributed inference solution, the MoAI Inference Framework, running on AMD GPUs.
At the event, Moreh presented a live demonstration of the GLM-5.1 large language model (LLM) powered by the MoAI Inference Framework on a system equipped with 32 AMD Instinct™ MI300X GPUs across four nodes. Visitors experienced the chatbot firsthand, evaluating its response speed and service quality while observing performance across a range of real-world use cases.
Unlike conventional demonstrations that simply run an AI model, Moreh’s showcase displayed key inference service metrics in real time, including GPU utilization, Tokens Per Second (TPS), Time To First Token (TTFT), and Time Per Output Token (TPOT). This enabled attendees to directly verify both inference performance and GPU resource efficiency in a production-like service environment.
Global AI industry leaders and enterprise customers attending the event expressed strong interest in the system’s fast response times and stable performance. In particular, the live deployment of the computationally demanding GLM-5.1 model on AMD GPUs at production-grade service levels received positive feedback from visitors.
Moreh’s MoAI Inference Framework is widely recognized as the world’s first commercially deployed distributed inference solution built for the AMD ecosystem. Its distributed inference and heterogeneous computing technologies are designed to dramatically reduce AI service costs, enabling broader adoption of AI worldwide. The technology addresses one of the industry’s biggest challenges-the rapidly rising infrastructure and service costs caused by increasingly larger AI models-by delivering a more efficient inference infrastructure.
Moreh CEO Gangwon Jo stated, “This event provided an opportunity for global customers to verify firsthand that top-tier inference performance can be achieved on AMD GPU environments,” and added “We will continue advancing our AI infrastructure software so enterprises can operate AI services as efficiently as possible, regardless of the underlying GPU platform.”
Moreh develops its own AI infrastructure engine and has expanded its end-to-end AI capabilities through its foundation LLM subsidiary, Motif Technologies, covering both AI infrastructure and foundation models. The company is also strengthening its presence in the global AI market through strategic partnerships with leading technology companies, including AMD and Tenstorrent.
SOURCE Moreh