Text-only HQQ-INT4 extraction of Gemma 4 E4B-it. Vision and audio branches are removed; the language_model submodule is preserved 1:1 and quantized with Half-Quadratic Quantization. Fits 8 GB consumer GPUs.

Property	Value
Base model	google/gemma-4-E4B-it
Quantization	INT4 weight-only
Approx. on-disk size	~9.2 GB
License	Gemma Terms of Use
Languages	English

Load (vLLM)

bash

vllm serve drawais/Gemma-4-E4B-text-HQQ-INT4 \
  --max-model-len 32768 \
  --gpu-memory-utilization 0.94

python

from vllm import LLM, SamplingParams
llm = LLM(model="drawais/Gemma-4-E4B-text-HQQ-INT4", max_model_len=32768)
print(llm.generate(["Hello!"], SamplingParams(max_tokens=128))[0].outputs[0].text)

Footprint

~9.2 GB on disk. Recommended VRAM: enough headroom for KV cache.

License & attribution

This artifact is a derivative work of google/gemma-4-E4B-it, released by its original authors under the Gemma Terms of Use.

This artifact is distributed under the same license. The full license text is included in LICENSE, and required attribution is in NOTICE.

License text: https://ai.google.dev/gemma/terms Source model: https://huggingface.co/google/gemma-4-E4B-it