Ask questions about an image privately
Run Gemma 4 12B in Ollama, submit a local image, and check its answer against visible details without sending the image to a hosted service.
Before you begin
- Difficulty
- Beginner
- Software
- Ollama
- Hardware
- Ollama lists a 7.6 GB package for Gemma 4 12B. Image input and context use extra memory, so use My Hardware only as a text-fit starting point.
Sources and files
Ollama Gemma 4 packages Official Ollama vision quick startChoose a model for this task
The steps below use the recommended model. Alternatives have their own package and command; open their model pages before switching.
The workflow
Install Ollama
Install the official app for your operating system and start its local service. Keep the image on the same computer as Ollama.
Download Gemma 4 12B
Ollama lists this 12B package with text and image input. Run it once to download the package before adding an image.
ollama run gemma4:12bPrepare a test image
Save a clear PNG or JPEG in your working directory. Start with a photo containing objects you can independently identify.
Ask about the image
Use Ollama’s documented image-path syntax with your local file. The example is adapted to the 12B Gemma 4 package; replace image.png with your file.
ollama run gemma4:12b ./image.png what is in this image?Check the answer
Confirm the reply refers to visible objects in the image. Ask a second question about a specific detail; do not treat a confident answer as proof if the detail is unclear.
Move to your real task
Try a diagram, screenshot, or photo that you are allowed to process. Keep prompts focused and reduce image size or other GPU load if memory becomes tight.
When it doesn’t go to plan
Confirm the selected package supports images, provide an existing local path, and follow the official vision syntax for your Ollama version.
Check the image is legible and the prompt asks about one detail. For speed or memory trouble, shorten the chat and close other GPU-heavy applications.
The model behind this workflow
Gemma 4 · 12BWill it run on your machine?
Save your machine to see a personalized rating and its reasoning.
Add my hardwareKeep exploring
Set up a private AI chat assistant
Install a local runtime, run Qwen3.5 9B, confirm responses, and know when to choose the smaller 4B package.
Build a private local coding assistant
Connect an open coding-capable model in Ollama to Cline, run a small repository task, and review the result locally.
Generate images on your own GPU
Use the official FLUX.2 Klein 4B ComfyUI template with exact model files, a first prompt, and an output check.