gemma-4-E4B-it-GGUF

gemma-4-E4B-it-GGUF

A 4-bit GGUF copy of Google's Gemma 4 E4B instruction model, packaged by Unsloth for running on your own machine. It reads text and images and answers in text.

Made by unsloth on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This is a GGUF copy of Gemma 4 E4B, the instruction-tuned model from Google DeepMind, packaged by Unsloth. It takes text and images and replies in text. The 4-bit weights are estimated at about 3.8 GB, so it can run on an ordinary computer.

Who it suits: Developers and hobbyists who want a small multimodal model on their own hardware, from laptops to phones. The card lists text generation, coding and reasoning as its strengths.

Industry | Who uses it | How they use it | Example
Software | App developers | Run a small local model with image input and function calling | Build a local assistant that answers questions about a screenshot
Research | ML practitioners | Test a compact instruction model in GGUF form | Compare its reasoning against a larger model
Local AI | Hobbyists | Chat with a model on a laptop without a cloud account | Load it in Unsloth Studio and ask coding questions
Multilingual work | Teams in many languages | Use one model that supports over 140 languages | Ask the same question in several languages

You can skip it if you need a hosted service or a model that runs only on large servers.

Where to use it:

• Unsloth Studio, where the card says Gemma 4 can run and be fine-tuned
• llama.cpp and other tools that read GGUF files
• Windows, macOS and Linux with Python, for downloading the files

Use case | Who | What they get back
Local chat | Hobbyists | Text answers from a model on your own machine
Image questions | Developers | Text answers about an image you send
Tool calling | Agent builders | Replies that can call functions, which the card says Gemma 4 supports natively

How to set it up: You need Python for the download and a program that can load GGUF files.

• Python on Windows, macOS or Linux
• Room for the 4-bit weights, estimated at about 3.8 GB
• A GGUF-capable tool such as Unsloth Studio or llama.cpp

1. Install Python if you do not have it.
2. Install the Hugging Face hub tool.
3. Download the model files.
4. Load them in your GGUF tool.

pip install -U huggingface_hub
hf download unsloth/gemma-4-E4B-it-GGUF

How to use it: The card shows no run command and points to a separate guide from Unsloth for running Gemma 4. In a chat tool, you write a message, attach an image if you like, and read the reply. The card notes a system role for setting the model's behavior.

Example: You attach a photo of a handwritten shopping list and ask the model to turn it into a typed list. It reads the image and answers with the items as text.

For newcomers and experts: A GGUF file is a model packed so ordinary computers can run it. A first small step is to download the files, open them in Unsloth Studio and ask a short question.

Gemma 4 comes in dense and mixture-of-experts variants. E2B and E4B are built for phones and laptops and handle 128K tokens of context. The larger sizes handle 256K. The card lists configurable thinking modes, native function calling and native system prompt support. Unsloth also points to its Dynamic 2.0 quantization benchmarks.

Download the GGUF files
Load them in a GGUF-capable tool
Send text and optional images
Read the text reply

Title: Context window in thousands of tokens
Small models such as E4B: 128 K tokens
Medium models: 256 K tokens

Good to know:

• The model has 7.5B parameters, and the 4-bit weights are about 3.8 GB by estimate.
• Audio input is listed for the E2B and E4B models, and video is among the supported inputs.
• The card says to re-download if you have an older copy, to get a newer chat template and llama.cpp fixes.
• License: apache-2.0.
• Cost to run: none named, you use your own hardware.
• Data it reaches: files downloaded from Hugging Face, and whatever text or images you give it.
• Not for: the card names no limits.

On Hugging Face, 13 November 2026

• Task: image text to text
• Parameters: 7.5B
• Weights: about 3.8 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 702,365
• Likes: 687
• License: apache-2.0
• Based on: google/gemma-4-E4B-it
• Quantized versions: 5
• Fine-tunes: 1
• Tags: gguf, gemma4, unsloth, gemma, google, image-text-to-text, conversational

Links

• Model: https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF
• Files: https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download unsloth/gemma-4-E4B-it-GGUF

Then follow the usage in its model card. Nothing in this kit runs by itself.