1 article tagged with this topic
Reddit user ran Qwen3-27B on 4 retired mining RTX 3060 Ti GPUs at 120 tokens/sec. The local AI hardware bar may be lower than you think.