mlx-community/gemma-4-26B-A4B-it-qat-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 [2, 5, 9, 11] -> [7, 9, 10, 11].

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`.

Sync chat template from google/gemma-4-31b-it (Google canonical, published 2026-07-09)

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

Correct BFCL + Capability: the AST checker could not match array-typed arguments

Card: put the comparison table in the same column order as every other OptiQ quant (ours, then uniform-4)

Move OptiQ sidecars under optiq/ so *.safetensors loaders (mlx-vlm, LM Studio) skip them

fix: audio_tower conv weights to MLX channel-last layout (full-VLM loader compat)

docs: OptiQ brand spelling

OptIQ mixed-precision 4-bit (QAT MoE), Capability 71.32 (+0.19 vs uniform-4)

initial commit