DiffGemma 26B-A4B-it — q4 (.dgq)
A quantized, self-contained .dgq pack of
DiffusionGemma 26B-A4B-it
(Gemma-4 26B-A4B MoE, discrete block diffusion) for the
diffgemma-mps Rust + Metal inference
engine on Apple Silicon. Text-only (v1).
Quantization — q4 profile
| Tensor class | Format | Bits/weight |
|---|---|---|
| MoE experts | q4_block (int4 affine, group 32) | 5.0 |
| Attention (q/k/v/o), dense FFN | bf16 | 16 |
| Embeddings, router, norms | bf16 | 16 |
| Self-conditioning MLP | q8_row | 8 |
~19 GiB of weights. Manifest version 1 (affine) — loads on every diffgemma-mps build.
Usage
diffgemma-mps download --repo mmastrac/diffgemma-26b-a4b-it-q4 -o model/diffgemma-26b-a4b-it-q4
diffgemma-mps -m model/diffgemma-26b-a4b-it-q4 chat
Build and run details: github.com/mmastrac/diffgemma.
License
Gemma Terms of Use. Derived from
google/diffusiongemma-26B-A4B-it; use is subject to the Gemma license.