Скачать или смотреть Running 120B Medical AI Models Locally | MINISFORUM MS-S1 MAX at AMD Booth

At the AMD booth at CES, MINISFORUM demonstrates a real-world on-device medical AI workflow using the MS-S1 MAX, built on AMD Ryzen™ AI Max+ 395.

The system is capable of running 120B-parameter large language models locally, delivering fast response speed and low latency without relying on cloud infrastructure.
Powered by an integrated NPU delivering up to 128 TOPS, combined with RDNA™ 3.5 integrated graphics and up to 96GB unified memory, MS-S1 MAX enables large-scale AI inference on a compact platform.

In this CES medical AI demo, doctors can:
Run local, private AI agents for healthcare workflows
Build domain-specific medical knowledge bases (RAG)
Analyze patient clinical records using large language models
Perform multi-agent reasoning across different medical roles
Upload and analyze CT and MRI images, including 3D imaging analysis, locally on the device

All AI workloads are processed entirely on-device, supporting data privacy and offline operation.
From the software side, the demo integrates a medical AI platform developed by an ISV partner, working together with AMD AI hardware capabilities.

📍 Live demonstration at the AMD booth at CES
🔒 Local AI inference for privacy-sensitive medical scenarios