Gemma 4 E2B Uncensored
HauhauCS’s Aggressive uncensored build of Google’s smallest Gemma 4 model: a 3.1 GB GGUF with reduced refusals for basic laptops. A community-modified checkpoint, not a Google release.
What do you want to do?
Choose a task. Get the software, files, settings, and steps to run it.
Open a guide for its software, hardware requirements, and setup steps.
At a glance
- Parameters
- 5.1B total
- Architecture
- Dense multimodal · community-modified checkpoint
- Context length
- 128,000 native (Google base model) · HauhauCS’s example uses 8,192
- License
- Gemma terms, as tagged on the repository · see the linked page
- Software
- llama.cpp, Ollama
Will it run on your machine?
Save your machine to see a personalized rating and its reasoning.
Add my hardwareBest for
- Private chat with fewer refusals on a laptop without a dedicated GPU.
- Try a local Hermes agent on modest hardware, as an experiment.
- Ask about images or audio by loading the extra file alongside the model.
Tradeoffs
- “0 refusals on 465 prompts” is HauhauCS’s own result, and its card notes Gemma 4 had less manual testing at long context than its other releases.
- The Aggressive build may still append short disclaimers, which HauhauCS does not count as refusals.
- The smallest Gemma 4 model: expect weaker reasoning and less reliable tool use than E4B or 12B.
- Not an official Google release. Google’s Gemma terms still apply.
Ways to run it
Ollama · Windows, macOS, Linux
IQ3_M is the only file in this repository with a standard quant name, so it is the only one Ollama can pull. Used by the Hermes guide.
ollama pull hf.co/HauhauCS/Gemma-4-E2B-Uncensored-HauhauCS-Aggressive:IQ3_Mllama.cpp · Windows, macOS, Linux
HauhauCS’s example settings. The Hermes agent guide on this page raises the context to 65,536.
llama-server --hf-repo HauhauCS/Gemma-4-E2B-Uncensored-HauhauCS-Aggressive --hf-file Gemma-4-E2B-Uncensored-HauhauCS-Aggressive-Q4_K_P.gguf --jinja -c 8192 -ngl 99What to expect
The pick for a basic laptop. Start with the 3.1 GB IQ3_M file in Ollama (ollama pull hf.co/HauhauCS/Gemma-4-E2B-Uncensored-HauhauCS-Aggressive:IQ3_M); it runs on a CPU.
Explore its uses
Sources and files
Primary sources and files behind this model page.
Keep exploring
Bonsai 2 · 27B
A compressed 27B-class reasoning model with publisher-provided GGUF packs and a dedicated llama.cpp fork for CUDA, Metal, and CPU.
DeepSeek V4.1 Flash
An open-weight model for reasoning, coding, agents, and image understanding. DeepSeek V4.1 Flash combines a 552B backbone with sparse activation and up to one million tokens of context. Choose the original checkpoint or a community uncensored variant below.
Qwen3.8
Explore 4 versions of Qwen3.8. Compare model sizes, packages, hardware requirements, and available setups.