Request for hardware support: Running Qwen models offline for quantum material research

[stas-creator]

Hello Qwen team,

I’m an independent researcher using llama.cpp to run Qwen-1.8B and Qwen2 models entirely offline for scientific analysis of nanoscale quantum structures in SiO₂. My goal is 100% local inference — no cloud, no data leakage.

My current machine (4 GB RAM, 10+ years old) limits me to small models. To scale to Qwen2-7B, I need a modest desktop (16+ GB RAM, Linux-compatible).

Since Qwen is designed for open and sovereign AI, I believe this use case aligns with your mission. If your team or Alibaba Cloud offers hardware grants or community support for offline AI research, I’d be honored to collaborate.

I’m happy to:

• Document the full setup publicly,
• Share benchmarks for low-RAM Qwen inference,
• Credit Qwen/ModelScope in all outputs.

Thank you for your incredible work on open LLMs.

— Stanislav Kravchenko, goodluckoll123@gmail.com