Google DeepMind
Build with Gemini Try Gemini

T5Gemma

Gemma

A collection of lightweight, state-of-the-art open models built from the same technology that powers our Gemini models*

Start building View docs

Models

VaultGemma: The world's most capable differentially private LLM

Our most advanced open models help developers create AI applications that run wherever users need them — from workstations to laptops and even phones.

Model family

Performance and benchmarks

Official variants

We’re exploring applications for Gemma as a code generator, in healthcare support roles, and even as a research tool in the animal kingdom.

Start building with Gemma

Supported by popular frameworks and platforms.

Hugging Face Transformers

Keras

Ollama

PyTorch

Gemma.cpp

JAX

MediaPipe

Google Cloud

Gemma Cookbook

A GitHub repository, collecting quickstart guides and code examples.

View cookbook

Developer events

Revisit past Developer Days, I/O sessions and other events where we’ve shared updates, and highlighted new opportunities for developers using our open models.

Building intelligent agents with Gemma 3

Explore the development of intelligent agents using Gemma models, with core components that facilitate agent creation, including capabilities for function calling, planning, and reasoning.

Welcome to Gemma 3

Discover the latest advancements in Gemma, Google's family of lightweight, state-of-the-art open models.

Deepdive into Gemma 3

Hear how the Gemma research team unveil the architecture, design principles, and innovations behind Google's family of lightweight, state-of-the-art open models.

A truly multilingual Gemma 3

Building multilingual AI applications is crucial for reaching global audiences, and varied language proficiency remains a top developer priority.

Explore the Gemmaverse

A vast ecosystem of community-created Gemma models and tools, ready to spark imagination and power innovation.

Start exploring

Large Language Models (LLMs), such as Gemma, may sometimes provide inaccurate or offensive content that doesn’t represent Google’s views.

Use discretion before relying on, publishing, or otherwise using content provided by LLMs.

Don’t rely on LLMs for medical, legal, financial, or other professional advice. Any content regarding those topics is provided for informational purposes only and is not a substitute for advice from a qualified professional.

T5Gemma

T5Gemma

A collection of encoder-decoder models that provide a strong quality-inference efficiency tradeoff

Read blog

Download T5Gemma

T5Gemma adapts pretrained decoder-only Gemma 2 models into an encoder-decoder architecture. These models are trained with either PrefixLM for strong generative performance or UL2 for high-quality contextual representations.

View technical report

Capabilities

  • Enhanced reasoning

    Dedicated encoder significantly boosts performance on tasks requiring deep context comprehension, such as math reasoning (GSM8K).

  • Flexible architecture

    Model adaptation techniques allows for flexible configurations, including "unbalanced" models where the encoder and decoder have different sizes.

  • High efficiency

    Superior quality-to-efficiency ratio without extensive compute requirements.

Model variants

  • Gemma 2 sizes

    Checkpoints based on the official Gemma 2 2B and 9B models, as well as the “unbalanced” 9B-2B checkpoint.

  • T5 sizes

    Small, Base, Large, and XL sizes following the T5 configuration, plus an additional model sized between T5 Large and T5 X.