gemma-3-27b-it-GGUF

gemma-3-27b-it-GGUF

Answers questions about text and images with Google's Gemma 3 27B instruction-tuned model, in GGUF files made by Unsloth for running on your own computer. It handles a 128K token context.

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 repository holds GGUF versions of gemma-3-27b-it, the 27B instruction-tuned model in Google DeepMind's Gemma 3 family. Unsloth made the quantized files. Gemma 3 takes text and images as input and writes text as output.

Who it suits: People who want a capable open model with image input on their own computer or cloud setup. The 4-bit weights are estimated at about 13.5 GB, so it needs a machine with room for that.

Industry | Who uses it | How they use it | Example
Software | Developers | Add a local model that reads text and pictures to an app | Describe the contents of a screenshot
Research | AI researchers | Test a 27B open model in quantized form | Compare answers from two file sizes
Education | Students | Practice question answering and summarizing on their own machine | Summarize a long article
Operations | Teams with their own hardware | Keep documents on their own systems | Ask questions about a long internal text

People who need the weights under a plain permissive license can skip it, because use falls under Google's Gemma terms.

Where to use it:

• Windows, macOS or Linux with Python, for downloading the files
• Tools that load GGUF files, such as llama.cpp or Ollama, which the card names as export targets
• Your own cloud setup, which the card mentions as a place to deploy Gemma 3

Use case | Who | What they get back
Question answering | Anyone | Written answers to questions about a text
Summarizing | Students and teams | A shorter version of a long document
Image understanding | Developers | A written analysis of what an image shows
Reasoning | Researchers | Step by step answers to problems

How to set it up:

• Python, for the Hugging Face hub tools
• A runner that loads GGUF files
• Disk space for the file you choose, with the 4-bit weights estimated at about 13.5 GB

1. Install the Hugging Face hub tools with pip.
2. Download the files with the hf command.
3. Load a GGUF file in your runner. The card points to Unsloth's guide for running Gemma 3 correctly.

pip install -U huggingface_hub
hf download unsloth/gemma-3-27b-it-GGUF

How to use it: Send the model text, or text with images, and read the text it writes back. The part of the card shown here gives no run command, so follow the runner you chose and the guide the card links. The details below come from the Gemma 3 model card.

Input | How it is handled
Text | Questions, instructions or a document you want shortened
Images | Resized to 896 x 896 and encoded to 256 tokens each
Context | The 27B size reads up to 128K tokens at once
Output | Generated text, such as an answer, an image analysis or a summary

Example: You give the model a photo and ask what it shows. It answers in text, describing the image content, as long as your runner supports image input.

For newcomers and experts: Start with plain text questions to check that the file loads. Once that works, try one image and a short question about it. Smaller laptops may need a smaller file than this one.

Gemma 3 is built from the same research and technology as the Gemini models. It covers more than 140 languages and has a 128K context window. At 27B, it sits at the top of the Gemma 3 size range, which also includes 1B, 4B and 12B.

Text or an image goes in
Images become 256 tokens each
The model reads the full context
It writes the answer as text

Good to know:

• Context: Up to 128K tokens for the 27B size.
• Images: Each image is resized to 896 x 896 and takes 256 tokens.
• Languages: The card says support covers over 140 languages. The record lists English.
• License: The Gemma license, which means Google's Gemma terms of use apply.
• Cost to run: none named. No API key or paid service appears in the source.
• Data it reaches: Files on your computer, plus the model download from Hugging Face.
• Not for: The source gives no forbidden uses in the part shown. The Gemma terms of use set the rules for use.

On Hugging Face, 13 November 2026

• Task: image text to text
• Library: transformers
• Parameters: 27B
• Weights: about 13.5 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 50,284
• Likes: 211
• License: gemma
• Languages: en
• Based on: google/gemma-3-27b-it
• Tags: transformers, gguf, gemma3, image-text-to-text, unsloth, gemma, google, imatrix, conversational

Links

• Model: https://huggingface.co/unsloth/gemma-3-27b-it-GGUF
• Files: https://huggingface.co/unsloth/gemma-3-27b-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-3-27b-it-GGUF

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