Uploaded October 2025 | Updated September 2026, 1 week ago

---

Experience the power of the MS-S1 MAX AI Cluster as it runs the massive DeepSeek-R1 671B (Q4) large language model locally, without cloud or internet access.

This demo showcases the full inference workflow on a 4-node cluster, from setup to reasoning output.

Cluster Specs: 4× MS-S1 MAX (Ryzen™ AI Max+ 395 “Strix Halo”)
Memory Usage: ~90GB per node during startup
Quantization: Q4 format
Mode: Local run, no embedded prompts

🛒 Order Now
👉 Enjoy $30 OFF with code S1MAX30
🌍 Global Store https://s.minisforum.com/3Kqve37
🇪🇺 EU Store https://s.minisforum.com/4gIlBsS
🇯🇵 JP Store https://s.minisforum.com/4gHpVs4
🇬🇧 UK Store https://s.minisforum.com/4pIAIGO
🇫🇷 FR Store https://s.minisforum.com/3IjUMyt
🇰🇷 KR Store https://s.minisforum.com/4nv1cKw

Chapters
00:00 4× MS-S1 MAX Cluster Overview
00:14 Memory Usage During Startup (~90GB per Node)
00:20 Network Traffic Between Nodes
00:29 Client Setup & Master Node Configuration
01:14 Running DeepSeek-R1 671B (Q4) on MS-S1 MAX Cluster

All responses are generated purely from the model’s internal memory, demonstrating distributed LLM inference on compact AI workstations.

#MINISFORUM #MSS1MAX #DeepSeek #AICluster #LLM #AIWorkstation #DeepSeekR1 #671B #TechDemo #LocalRun #ClusterRun #AIComputing