[enhancement]: Qwen2.5-VL GGUF text encoder (Qwen-Image) #9732

Is there an existing issue for this?

• I have searched the existing issues

What should this feature add?

Load GGUF builds of the Qwen2.5-VL text encoder used by Qwen-Image. The Qwen-Image transformer already loads from
GGUF, but the encoder only loads as a folder, bf16/fp8 single file or nvfp4; there is no GGUF config for
QwenVLEncoder.

Alternatives

Use the nvfp4 or fp8 encoder.

Additional Content

• Demand: GGUF builds of Qwen2.5-VL-7B are common (e.g. unsloth/Qwen2.5-VL-7B-Instruct-GGUF), and Qwen-Image
  encoder variants are published as GGUF (e.g. pottokao/Qwen-Image-2.1-Text-Encoder-Heretic-GGUF, ~225k).
• The Qwen3-VL GGUF encoder loader (language tower only, mmproj refused) is the closest template.
• Qwen-Image 2.1 itself is tracked in [enhancement]: add Qwen-Image-2.1 support #9594.

The full matrix of what loads today is in #9726 (Model Format Support page).