mlx-community/Qwythos-9B-v2-OptiQ-4bit

Update kv_config.json: re-measured with `optiq kv-cache` float32 sensitivity (the previous config was ranked on bfloat16 rounding noise). 8-bit layers [15] -> [7].

Add kv_config.json: per-layer KV cache precision from `optiq kv-cache` (target 4.5 bits). Use with `optiq serve --kv-config kv_config.json`.

index: register the bf16 vision tower so stock VLM loaders (mlx-vlm, oMLX, LM Studio) can find it

OptiQ mixed-precision 4/8-bit quant (5.211 bpw) with bf16 vision sidecar + MTP head

initial commit