Sponsored by Byond Boundrys Consulting - Empowering Ideas, Delivering Results
Google DeepMind Intermediate

Gemma 2

Gemma 2 is Google's open-weight text family with 2B, 9B and 27B variants for generation, summarization, question answering and reasoning.

Language Models (LLMs)Text Open Source
In plain English

What is this model and why does it matter?

Gemma 2 is Google's open-weight text model family with 2B, 9B and 27B sizes for generation, summarization, question answering and reasoning.

Local text AIFine-tuningResearchSummarizationQuestion answeringEducation
Model overview

Gemma 2: features, use cases and important details

Gemma 2 is Google’s second-generation Gemma family of lightweight text models. Google released the 9B and 27B sizes on June 27, 2024 and added the smaller 2B size on July 31, 2024.

What is in the Gemma 2 family?

The official model card covers 2B, 9B and 27B sizes, with pretrained and instruction-tuned variants. Gemma 2 is text-to-text, English-focused and distributed as open weights under Google’s Gemma Terms of Use.

Where Gemma 2 fits

The family is suitable for question answering, summarization, reasoning, education and custom fine-tuning. The different sizes allow developers to trade hardware requirements against model capability, from relatively lightweight 2B deployments to the much larger 27B model.

Important distinction

Gemma 2 is not a Gemini API model and there is no single provider token price for the family. This page should act as an overview and link to the exact Gemma 2 2B IT, 9B IT and 27B IT pages for checkpoint-level details.

Get started

How to use this model

  1. Choose a 2B, 9B or 27B checkpoint and decide between pretrained or instruction-tuned weights.
  2. Accept the Gemma terms.
  3. Download the official model.
  4. Run with a supported framework or deploy through a cloud environment.
  5. Evaluate the selected size on your workload.
Copy and try

Example prompts

  • Summarize this document into five key points.
  • Explain this scientific concept for a high-school student.
  • Rewrite this paragraph in clearer English.
Capabilities

What it can do

  • Text generation
  • Question answering
  • Summarization
  • Reasoning
  • Fine-tuning
Best for

Practical use cases

  • Local assistants
  • Education
  • Research
  • Document summarization
  • Custom fine-tuned models
Pricing

What does it cost?

Open-weight Gemma 2 family under Google's Gemma terms; no single provider token price applies to the downloadable checkpoints.

Simple summaryWeights are downloadable under the Gemma terms. Costs depend on the selected size, hardware and deployment provider.

What stands out

  • Three model sizes
  • Official open weights
  • Good local deployment options
  • Clear Google model documentation

Things to consider

  • English-focused
  • 8K context is short by current standards
  • Gemma terms are not the same as a permissive open-source license
Limitations

Important restrictions and trade-offs

  • Can produce inaccurate outputs
  • Family page spans multiple sizes and pretrained/instruction variants
  • No single token price or hosted API identity applies
SimplifyAITools verdict

Our editorial take

A useful family page for comparing Gemma 2 sizes, especially for self-hosting and fine-tuning; link users to exact 2B, 9B and 27B profiles for checkpoint-specific details.

References

Primary sources

  1. Open source 1 ↗
  2. Open source 2 ↗
  3. Open source 3 ↗
Keep exploring

Related AI models