Google models

Gemini is Google’s most capable AI model family, designed to be multimodal, efficient, and highly capable at a wide variety of tasks.

​

Using the API

Use Chat completion API to call Gemini models. See the pricing page for pricing details.

​

Gemini models comparison

​

Important notes

​

Long context handling

• All models support up to 1M tokens context window
• Different pricing tiers apply based on input length:
  • Standard rate for inputs less than 128k tokens
  • Higher rate for inputs more than 128k tokens
• Please check the pricing page for detailed pricing information

​

Vision capabilities

• Gemini 2.5 Pro Preview and Gemini 1.5 Pro support multimodal inputs including images
• Can analyze complex visual information and provide detailed descriptions
• Vision support is not available in Flash or Flash-Lite models

​

Model strengths

• Gemini 2.5 Pro Preview:

  • Advanced reasoning and analysis
  • Multimodal understanding
  • Complex task handling
  • Code generation and analysis

• Gemini 2.5 Flash:

  • Fast response times
  • Cost-effective for simple tasks
  • Efficient token usage
  • Good for high-throughput applications

• Gemini 2.5 Flash-Lite:

  • Lightweight and efficient
  • Ideal for quick responses
  • Budget-friendly for simple tasks

• Gemini 2.0 Flash:

  • Cost-effective for basic tasks
  • Reliable for lightweight applications
  • Efficient token usage

• Gemini 1.5 Pro:

  • Previous generation advanced model
  • Multimodal capabilities
  • Complex reasoning and analysis

​

Official documentation

• Gemini API Overview
• Vision Guide
• Safety & Responsible AI