LightweightOpen SourceGoogle

Gemma 3 27B

Gemma 3 27B is Google's open-source lightweight model with text and image capabilities, featuring a 128K token context window and multimodal understanding.

Context 128K
Tier Lightweight
License Open Source
Modalities text, image
Input from
$0.020 / 1M tokens
across 4 providers

API Pricing

Cheapest on Together AI 81% below avg
ProviderInput / 1MOutput / 1MSpeedTTFTUpdated
$0.020$0.04017.4 t/s386ms4/14/2026
$0.080$0.16017.4 t/s386ms4/4/2026
$0.080$0.16017.4 t/s386ms4/14/2026
$0.120$0.19017.4 t/s386ms4/14/2026
$0.230$0.38017.4 t/s386ms4/14/2026

Prices updated daily. Last check: 4/14/2026

Model Details

General

Creator
Google
Family
Gemma
Tier
Lightweight
Context Window
128K
Modalities
Text, Image

Capabilities

Tool Calling
No
Open Source
Yes
Subtypes
Chat Completion
Aliases
gemma-3-27b-it, gemma-3n-e4b-it

Strengths & Limitations

  • Open-source with accessible model weights for local deployment
  • Multimodal support for both text and image inputs
  • 128K token context window for processing lengthy documents
  • Fast inference speed at 39.07 tokens per second
  • Low time to first token at 308ms for responsive applications
  • Lightweight 27B parameter size reduces computational requirements
  • Created by Google with established model development expertise
  • No function calling or tool use capabilities
  • Smaller parameter count than flagship models in other families
  • Limited to chat completion format without structured output modes
  • Fewer advanced API features compared to proprietary alternatives
  • May have reduced reasoning capabilities versus larger models

Key Features

128K token context window
Text and image input processing
Open-source model weights
Chat completion interface
Streaming response support
Multi-turn conversation handling
Document analysis capabilities
Vision-language understanding

About Gemma 3 27B

Gemma 3 27B is Google's open-source lightweight model in the Gemma family, designed to provide capable performance with relatively modest computational requirements. As a 27 billion parameter model, it sits in the lightweight tier while offering both text and image processing capabilities through its multimodal architecture. The model features a 128K token context window and supports both text generation and image understanding tasks. With benchmark performance showing 39.07 output tokens per second and a time to first token of 308ms, it demonstrates efficient inference characteristics. The model is available under an open-source license, allowing developers to access model weights and run inference locally or modify the model for specific use cases. Gemma 3 27B targets applications where developers need multimodal capabilities without the computational overhead of larger frontier models. Its open-source nature makes it particularly suitable for organizations requiring on-premises deployment, model customization, or cost-effective scaling of text and vision tasks.

Common Use Cases

Gemma 3 27B is well-suited for organizations needing cost-effective multimodal AI capabilities with the flexibility of open-source deployment. Its combination of text and image processing makes it valuable for document analysis, content moderation, educational applications, and customer support systems that handle both text and visual inputs. The model's lightweight nature and fast inference speed make it practical for high-volume applications where response time matters, while its open-source license enables customization for domain-specific tasks, on-premises deployment for data privacy requirements, and integration into existing infrastructure without vendor dependencies.

Frequently Asked Questions

How much does Gemma 3 27B cost per million tokens?

Gemma 3 27B pricing varies by provider and deployment method. As an open-source model, you can run it locally without API costs, or use hosted providers with different pricing structures. Check the pricing table above for current rates across all providers offering Gemma 3 27B.

What is Gemma 3 27B best used for?

Gemma 3 27B excels at multimodal tasks requiring both text and image understanding, such as document analysis, content moderation, and educational applications. Its lightweight design and fast inference make it suitable for high-volume deployments, while its open-source nature enables customization and on-premises deployment for organizations with specific requirements.

Can I run Gemma 3 27B locally or do I need an API?

Gemma 3 27B is open-source, so you can download the model weights and run it locally on your own infrastructure. This gives you full control over deployment, data privacy, and costs. Alternatively, you can use hosted API providers for easier setup without managing infrastructure yourself.