AnalysisAI ModelsSeptember 6, 2026

Reddit user runs Qwen 3.8 27B locally on dual R9700 GPUs with vLLM

A LocalLLaMA user reports running Qwen 3.8 27B on two R9700 GPUs with 64 GB DDR5 using vLLM Radiance / R9V, upgrading from an Intel Arc B580 and 32 GB DDR5. They cite the 5090's roughly 3x price as the reason for the R9700 route.

1 source

More stories today

Open the live feed