Gemma 4 26B A4B (mixture-of-experts, ~4B active) instruction-tuned, quantized to the Apertura .apml bundle format: MLX-affine 4-bit (group size 64, 8-bit embeddings), tokenizer.json + chat_template.jinja included, runtime mlx, architecture gemma4. Post-training quantization of the bf16 release (this family has no QAT variant).

Consumed by Apertura — an Objective-C++ transformer engine + macOS chat app on MLX. Export gate: --verify-bundle (bundle reload == in-memory quantization, argmax-identical).

Gemma is provided under and subject to the Gemma Terms of Use. By downloading this model you agree to those terms, including the Gemma Prohibited Use Policy. This repository redistributes a quantized Model Derivative of google/gemma-4-26b-a4b-it; the same terms and use restrictions apply to it and to any further derivatives.