Sign up

AtomicChat/gemma-4-26B-A4B-it-GGUF

🤗 On Hugging Facetext-generationapache-2.0364 GBGGUF✓ HF checksums availableupdated today
Magnet

Base model: google/gemma-4-26B-A4B-it

Gemma 4 26B A4B, self-quantized to GGUF by Atomic Chat. Built straight from Google's original weights with a per-tensor importance matrix, so this is not a repack of somebody else's files. Runs fully offline.

Highlights

  • 25.2B total / 3.8B active per token parameters: the weights this repo quantizes.
  • Context length: 256K tokens, as published by Google.
  • 30 layers: Mixture-of-Experts, hybrid sliding-window (1024) and global attention.
  • Modalities: the base model handles Text, Image; this repo ships text-only quants, it carries no vision projector.
  • Full imatrix ladder: every quant is calibrated with an importance matrix, published here alongside the quants.
  • Reasoning: All models in the family are designed as highly capable reasoners, with configurable thinking modes.
  • Diverse & Efficient Architectures: Offers Dense and Mixture-of-Experts (MoE) variants of different sizes for scalable deployment.
[!NOTE]
These GGUFs are self-quantized from the original weights, not a repack. The importance matrix keeps low-bit quants closer to the full-precision model.
[!IMPORTANT]
Always pass --jinja so the Gemma 4 26B A4B chat template is applied. Without it the model can emit malformed turns.

Model Overview

| Property | Value |

|---|---|

| Base model | google/gemma-4-26B-A4B-it |

| Parameters | 25.2B total / 3.8B active per token |

| Layers | 30 |

| Experts | 128 routed (top-8) |

| Sliding window | 1024 tokens |

| Context length | 256K tokens |

| Vocabulary | 262K |

| Modalities | Text, Image in the base model; text only in this repo, it ships no vision projector |

| Architecture | Mixture-of-Experts, 128 experts (top-8), hybrid sliding-window (1024) and global attention, 16 attention heads over 8 KV heads, Gemma4ForConditionalGeneration |

| This repo | GGUF quants (imatrix); the importance matrix is published here as imatrix-coding.gguf. Quants: Q2_K, IQ3_M, Q3_K_M, Q3_K_L, IQ4_XS, Q4_K_S, Q4_K_M, UD-Q4_K_XL, Q5_K_S, Q5_K_M, Q6_K, Q8_0 |

Scores are Google's published results for the base google/gemma-4-26B-A4B-it, not our own measurements. Quantization preserves the large majority of this; Q4_K_M and up stay close to full precision.

Choosing a quant

| Quant | Size | Notes |

|---|---|---|

| Q2_K | 10.6 GB | Smallest K-quant. Minimal RAM, clear quality drop. |

| IQ3_M | 12.4 GB | Beats Q3 at a similar size thanks to imatrix. Best low-RAM pick. |

| Q3_K_M | 13.3 GB | Low quality but usable. |

| Q3_K_L | 13.8 GB | A step above Q3_K_M. |

| IQ4_XS | 13.9 GB | Excellent quality for size. Recommended low-bit. |

| Q4_K_S | 15.5 GB | Compact 4-bit, fast. |

| Q4_K_M | 16.8 GB | Recommended default. Best balance of size, speed and quality. |

| UD-Q4_K_XL | 17.0 GB | Dynamic. Embeddings and output kept at Q8_0 for higher quality at a Q4 footprint. |

| Q5_K_S | 18.0 GB | Higher quality, slightly more compact than Q5_K_M. |

| Q5_K_M | 19.1 GB | Higher quality, low loss. |

| Q6_K | 22.6 GB | Near lossless, noticeably lighter than Q8_0. |

| Q8_0 | 26.9 GB | Effectively lossless, reference quality. |

[!TIP]
Pick the largest file that fits your (V)RAM with room for context. Q4_K_M or UD-Q4_K_XL is the sweet spot for most setups; Q6_K or Q8_0 for maximum fidelity.

Get started

Run Gemma 4 26B A4B locally with:

  • Atomic Chat: the easiest path. Open the app, search AtomicChat/gemma-4-26B-A4B-it-GGUF, pick a quant, hit Use this model.
  • llama.cpp: llama-server -hf AtomicChat/gemma-4-26B-A4B-it-GGUF:Q4_K_M --jinja -c 8192
  • Ollama: ollama run hf.co/AtomicChat/gemma-4-26B-A4B-it-GGUF:Q4_K_M
  • LM Studio / Jan: search the repo id, download any quant.

Best practices

| Parameter | Value |

|---|---|

| temperature | 1.0 |

| top_p | 0.95 |

| top_k | 64 |

Google's recommended sampling configuration for google/gemma-4-26B-A4B-it.

Run in llama.cpp

git clone https://github.com/ggml-org/llama.cpp
cmake llama.cpp -B llama.cpp/build -DBUILD_SHARED_LIBS=OFF -DGGML_CUDA=ON
cmake --build llama.cpp/build --config Release -j --target llama-cli llama-server
./llama.cpp/build/bin/llama-server \
    -hf AtomicChat/gemma-4-26B-A4B-it-GGUF:Q4_K_M \
    --jinja -ngl 99 -c 8192 -fa on

How these were made

1. Download google/gemma-4-26B-A4B-it (original weights).

2. Convert to f16 GGUF with llama.cpp.

3. Build an importance matrix over our calibration corpus, published here as imatrix-coding.gguf.

4. Quantize the ladder with --imatrix.

5. UD-Q4_K_XL additionally pins the token-embedding and output tensors to Q8_0.

License

Original model by Google, released under the Apache 2.0 license. Full terms: Apache 2.0. Quantized by Atomic Chat.