Introducing Gemma 4 12B: a unified, encoder-free multimodal model

Today we’re introducing Gemma 4 12B, our latest model designed to bring agentic multimodal intelligence directly to laptops. Gemma 4 12B bridges the gap between our edge-friendly E4B and our more advanced 26B Mixture of Experts (MoE), combining powerful features in a reduced storage footprint. It’s also our first mid-size model to feature native audio inputs.

Thanks to the developer community, the Gemma 4 models have now reached the 150 million download mark. They’ve built everything from wearable robotic arms for physical support to enterprise-grade AI security. We’re excited to see what you build with this latest addition.

Here’s an overview of what makes Gemma 4 12B unique:

• Novel unified architecture: No multimodal encoders. The image and audio inputs flow directly into the LLM backbone.
• Extended reasoning: Benchmark performance close to our 26B model, enabling powerful multi-stage reasoning and agent workflows.
• Laptop ready: Small enough to run locally with just 16GB of VRAM…