Chinese supercomputer manufacturer Sugon has unveiled a new mobile workstation designed for AI tasks, featuring a 64-thread domestic CPU and a discrete GPU with 16GB of VRAM. This combination aims to deliver cloud-level AI inference performance locally, processing up to 50 tokens per second on 35B MoE models. The workstation is notably thin, measuring just 16.9 mm, and is expected to offer significantly faster AI performance compared to current market offerings. AI
IMPACT This workstation could enable more powerful local AI inference, potentially reducing reliance on cloud services for certain tasks.
RANK_REASON This is a product announcement for a workstation, not a core AI model release or research milestone.
- AMD
- C86-1G
- C86-4G
- C86-5G
- China
- Dell
- Hewlett Packard Enterprise
- Hygon Information Technology
- Sugon
- Zenodo
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →