---
license: apache-2.0
base_model: google/gemma-4-26B-A4B-it
tags: [axom, gguf, llama.cpp, lora]
---
# Axom X1-Ultra — X1-Ultra (26B MoE Q4)

Files an Axom install downloads for the **X1-Ultra** tier (`axom init`), mirrored here so no
Hugging Face gate stands between a user and the model. Gemma 4 is Apache 2.0.

| file | what | sha256 |
|---|---|---|
| `gemma-4-26B_q4_0-it.gguf` | chassis: the official Google QAT GGUF, unmodified | `3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d` |
| `ops_26b_v1-lora-f16.gguf` | ops leaf (ops_26b_v1) | `54e0929e6fa77c2e8fb1b13257e4d00645731f9b73810d14f658862c82433fe1` |

The chassis is the official `google/gemma-4-26B-A4B-it-qat-q4_0-gguf` file, byte for byte. The leaf is an
attention-only LoRA pressed by Axom Labs on the QAT-unquantized base of the same file. Runtime and docs:
Axom-CLI. Install: `curl -sSf https://get.axomlabs.ai/install.sh | sh`.