incept5/Qwen3.8-27B-MLX-4bit

Verification: add reproducible test image, inference snippet, and results

Add smoke-test image

Add smoke-test results (text + vision) to model card

Add MLX 4-bit weights (converted with mlx-vlm 0.6.13)

Add model card (original Qwen3.8-27B content + MLX 4-bit notes)

initial commit