Open-weight & owned stack · Quantization efficiency
Qwen runs at sixty tokens on dual mining APUs
A user reported running Qwen3.6-35B-A3B at 60 tokens per second with 64K context across two BC-250 APUs costing about $300.
Read the original at reddit.comOpens the publisher's site in a new tab