Gemma 3
Gemma 3 with vision support. 12B is the sweet spot for single-GPU multimodal chat.
What Gemma 3 is for
Gemma 3 takes images as well as text. Google shipped it in 2025-03 with 5 sizes (270M, 1B, 4B, 12B, 27B), 128K of context and a Gemma License licence. Running a vision-language model locally costs more memory than a text model of the same parameter count, because the image encoder and the projected image tokens both occupy the context. The catalogue records its strength as single-GPU multimodal chat.
Sizes and memory
The 1B needs about 1 GB of weights at Q4_K_M and the 27B about 17 GB, a spread of 16 GB across the family. The catalogue's default pick is 12B. Recommended VRAM adds roughly half again on top of the weights for the KV cache and the runtime.
| Size | Weights at Q4_K_M | Recommended VRAM | Smallest card in our guide that holds it |
|---|---|---|---|
| 1B | 1 GB | 2 GB | NVIDIA Jetson Orin Nano 8GB |
| 4B | 3 GB | 5 GB | NVIDIA Jetson Orin Nano 8GB |
| 12B | 8 GB | 12 GB | NVIDIA RTX 3060 12GB |
| 27B | 17 GB | 26 GB | NVIDIA RTX 3090 |
Hardware that runs Gemma 3
Every one of the 19 cards with dedicated memory in our hardware guide holds at least one size of Gemma 3 at Q4_K_M. The cheapest that does is the NVIDIA Jetson Orin Nano 8GB at about $199, running the 4B.
The largest size any of them holds is the 27B at 17 GB, on the NVIDIA RTX 3090 (24 GB at 936 GB/s). Bandwidth, not capacity, sets the speed once a size fits, so a card further down the list with faster memory will generate more tokens per second on the same weights than a larger, slower one.
What people use it for
The catalogue lists 4 recorded uses for Gemma 3: chat, vision, multilingual, edge. It is tagged chat, vision, multilingual.
Where Gemma 3 sits in the Gemma family
Our catalogue holds 3 Gemma entries. Ordered by release date, Gemma 3 (2025-03) is the 2nd of them, and Gemma 4 (2026-05) is newer.
It ships under the Gemma License, which is neither a standard open-source licence nor a closed one. Read the terms before shipping anything commercial on top of it.
Context window: 128K
128K tokens is roughly 96,000 words at the usual ratio, which is a small repository or a whole book. Two cautions come with a window this size: the KV cache at full length can rival the weights for memory, and models rarely use the far end of their advertised context as well as they use the near end. Treat it as headroom, not as a promise.
How to run Gemma 3 locally
Our catalogue does not record registry tags, so look the current tag up in the Ollama library or on Hugging Face before pasting these. The commands below are the shape of the workflow, not a copy-and-paste recipe.
Option 1: Ollama (simplest)
# find the tag for Gemma 3 at ollama.com/library
ollama run <tag>Option 2: Mullama (production)
mullama pull <tag>
mullama run <tag>Option 3: llama.cpp (CLI)
# download a GGUF from Hugging Face, searching for "Gemma 3 GGUF"
./llama-cli -m model.Q4_K_M.gguf -p "Hello, AI!"Option 4: Python with Mullama or llama-cpp-python
from mullama import Model, Context
model = Model.load("model.Q4_K_M.gguf", n_gpu_layers=99)
ctx = Context(model, n_ctx=4096)
print(ctx.generate("Hello, AI!", 256))Sources
Model record from src/data/models.json (2026-06-29); card capacities from src/data/gpus.json (2026-06-29). Upstream: Curated from ollama.com/library + community benchmarks + paperswithcode + huggingface.