Quickstart¶

Gemma is a family of lightweight, state-of-the-art open models built from the same research and technology used to create the Gemini models. Developed by Google DeepMind and other teams across Google, Gemma is named after the Latin gemma, meaning precious stone. The Gemma model weights are supported by developer tools that promote innovation, collaboration, and the responsible use of artificial intelligence (AI).

The latest in the Gemma family is Gemma 4 featuring multimodal (text, image, audio) inputs.

Choose a model¶

Gemma 4 ships in five sizes. They share a tokenizer and prompt format, so you can start small and move up without rewriting anything.

Run your first inference¶

from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="google/gemma-4-12B-it",
    device_map="auto"
)
output = pipe("Explain quantum computing simply:", max_new_tokens=150)
print(output[0]["generated_text"])

llama-cli -hf google/gemma-4-12B-it-qat-q4_0-gguf --prompt "Write a poem about the Kraken."

ollama run gemma4:12b "Explain quantum computing simply:"

mlx_lm.generate --model mlx-community/gemma-4-12B-it-qat-4bit \
    --prompt "Explain quantum computing simply:"

vllm serve google/gemma-4-26B-A4B-it

lms chat google/gemma-4-12b

npx skills add google-gemma/gemma-skills --skill gemma-dev --global

Providers¶

If you want to run Gemma 4 on a different stack, you can choose from the following:

• Ollama
• lc llama.cpp
• LM Studio
• MLX
• Li LiteRT
• Transformers
• Keras
• Un Unsloth
• vLLM
• SG SGLang
• Gemini API
• Google Cloud
• All ways to run →

Get the weights¶

Every size ships full-precision weights plus Quantization-Aware Training (QAT) checkpoints, which hold up far better at 4-bit than a generic post-training quant.

Checkpoint names follow google/gemma-4-{SIZE}-it{SUFFIX} — pick the suffix your engine wants:

What's next¶