Request: AVX-only (no AVX2) GGUF builds for Qwen2.5-VL and Qwen3.5

Hello Qwen team,

Thank you for the excellent Qwen2.5-VL and Qwen3.5 model families — they are truly impressive.

I am writing on behalf of users with older (but still functional) hardware. I have a 2011 laptop with Intel Core i7-2720QM (Sandy Bridge, AVX but NO AVX2), 16 GB RAM, SSD, running Windows 11. This machine is still perfectly usable for daily tasks, but cannot run modern local LLMs because:

1. Current GGUF builds from popular quantizers (bartowski, mradermacher, huihui-ai) require AVX2.
2. Older runtimes (e.g. llama.cpp bundled in LM Studio 0.2.10 AVX-Beta) do not recognize the qwen2vl and qwen35 architectures.

Could you please consider:

• Publishing official AVX-only GGUF builds of Qwen2.5-VL-7B and Qwen3.5-9B (Q4_K_M / Q6_K)?
• Or coordinating with the llama.cpp team to backport qwen2vl/qwen35 architecture support to AVX-only builds?

There is a meaningful community of users with Sandy Bridge / Ivy Bridge CPUs (2011–2012 era) who would greatly benefit from running Qwen models locally. Many of us prefer local inference for privacy reasons and cannot always afford newer hardware.

Thank you for your time and your wonderful work.

Best regards, Valery