Paper page - Gemma 4 Technical Report

Source: https://huggingface.co/papers/2607.02770 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

Gemma 4 introduces efficient, multimodal language models with diverse architectures, enhanced reasoning capabilities, and improved performance across various tasks.

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 andMixture-of-Experts architectures, ranging from 2.3B to 31B parameters. Alongside improved vision andaudio encodersfor all model sizes, we propose a unified,encoder-free architecturefor our 12B model, which ingests raw audio and image patches. Furthermore, we integrate athinking mode, enabling Gemma models to generate reasoning traces prior to responding. We improve inference speed, memory, and compute efficiency, as well aslong-context abilitiesthrough 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.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2607\.02770

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.02770 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.02770 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.02770 in a Space README.md to link it from this page.

Collections including this paper2