Qwen3.8-Flash-Next-GSQ-RCO (IQ3_S): ~20-30 tok/sec decode & 300-90k tok/sec prefill on 12GB VRAM + 32GB RAM + NVME

🌐 链接: https://www.reddit.com/r/LocalLLaMA/comments/1x1mzsg/qwen38flashnextgsqrco_iq3_s_2030_toksec_decode/
🏷️ 分组: Reddit｜AI·自托管·VPS
🕒 时间: 2026-10-09 22:34:14
Reddit
From the LocalLLaMA community on Reddit: Qwen3.8-Flash-Next-GSQ-RCO (IQ3_S): ~20-30 tok/sec decode & 300-90k tok/sec prefill on…
From the LocalLLaMA community on Reddit: Qwen3.8-Flash-Next-GSQ-RCO (IQ3_S): ~20-30 tok/sec decode & 300-90k tok/sec prefill on…
Explore this post and more from the LocalLLaMA community