CoolFace
Modelpublic

noctrex/gemma-4-26B-A4B-it-MXFP4_MOE-GGUF

sourceHugging Faceupdated 5mo agoView on Hugging Face
22likes2kdownloads
Model Card

These are MXFP4 quantizations of the model gemma-4-26B-A4B-it

Quick Start

  1. 1.Download the latest release of llama.cpp.
  2. 2.Download your preferred model variant from below.
  3. 3.For the mmproj file, it is recommended to use the F32 version for the best visual processing results. F32 > BF16 > F16

Which version should I choose?

All variants use MXFP4 for the MoE (Mixture of Experts) weights to keep the model efficient. The difference lies in how the remaining tensors are handled:

VariantQualityPerformanceSizeRecommendation
BF16⭐⭐⭐Variable*15.80GiBBest for maximum accuracy; original unquantized weights.
F16⭐⭐Fast15.80GiBGreat alternative if BF16 is slow on your hardware.
Q8⭐Fastest14.36GiBBalanced performance and memory usage.

Note: On some older architectures, BF16 may be slower than F16. Check that your GPU supports native BF16

Read the guide from unsloth in order to set up the model's recommended settings:

The official chat template has been updated from Google. If you do not want to download the model again, you can just tell llama.cpp to use the new chat template.