Gemma 4 Technical Report
This technical report presents incremental improvements in compute efficiency and reasoning for open-weight multimodal models, targeting researchers and developers needing accessible yet capable models.
Gemma 4 introduces a new generation of open-weight, natively multimodal language models with dense and MoE architectures (2.3B–31B parameters), featuring a unified encoder-free design for the 12B model and a thinking mode for reasoning. It achieves significant performance gains across STEM, multimodal, and long-context benchmarks, rivaling larger open models.
We introduce Gemma 4, a new generation of open-weight, natively multimodal language models in the Gemma model family. Designed to advance compute efficiency and reasoning, the Gemma 4 model suite features dense and Mixture-of-Experts architectures, ranging from 2.3B to 31B parameters. Alongside improved vision and audio encoders for all model sizes, we propose a unified, encoder-free architecture for our 12B model, which ingests raw audio and image patches. Furthermore, we integrate a thinking mode, enabling Gemma models to generate reasoning traces prior to responding. We improve inference speed, memory, and compute efficiency, as well as long-context abilities through critical design choices. Gemma 4 establishes a leap in performance across STEM, multimodal, and long-context benchmarks, and rivals larger, frontier open models in human-rated tasks.