Qwen 3.6 35B A3B: 131K context + vision on 6GB VRAM
英文标题：Qwen 3.6 35B A3B: 131K context + vision on 6GB VRAM
Running HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive Q4_K_M with llama.cpp at ~600 tok/s prefill and 23 tok/s decode, 131k context window, Q8 KV cache - on an RTX 2060 6GB + 32GB DDR4 RAM. Speeds start at ~600 tok/s prefill / 23 tok/s decode on an empty KV cache. As context grows they set
打开原文相关 2 · 机会 0
中文
Running HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive Q4_K_M with llama.cpp at ~600 tok/s prefill and 23 tok/s decode, 131k context window, Q8 KV cache - on an RTX 2060 6GB + 32GB DDR4 RAM. Speeds start at ~600 tok/s prefill / 23 tok/s decode on an empty KV cache. As context grows they set
中文全文翻译正在补齐，可先查看下方英文原文。
English / 原文
Running HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive Q4_K_M with llama.cpp at ~600 tok/s prefill and 23 tok/s decode, 131k context window, Q8 KV cache - on an RTX 2060 6GB + 32GB DDR4 RAM. Speeds start at ~600 tok/s prefill / 23 tok/s decode on an empty KV cache. As context grows they set