カオスマップに戻る
Gemini (Google)

Gemini (Google)

Overview

Gemini Series is Google DeepMind's family of multimodal foundation models. Developed by the merged Google DeepMind organization (Google Brain + DeepMind), Gemini natively handles text, images, audio, video, and code. As of 2026, the main models are Gemini 3.1 Pro (latest flagship) and Gemini 2.5 Pro / 2.5 Flash, deeply integrated with Google Workspace and Vertex AI.

Latest Models: Gemini 3.1 Pro / 2.5 Pro

Lineup

ModelUse CaseAPI Price (Input/Output per 1M tokens)
Gemini 3.1 ProLatest flagship$2 / $12
Gemini 2.5 ProHigh performance, built-in thinking$1.25 / $10
Gemini 2.5 FlashLow-cost, fast$0.15 / $0.60

Key Features

1. Native Multimodal

Single model handles text, images, audio, video, and code natively. Strong performance on video temporal understanding, PDFs, and codebase comprehension.

2. 1M Token Context (2M Coming)

Gemini 2.5 Pro ships with a 1M token context window. Process entire books, codebases, or long videos at once. 2M tokens is coming.

3. Thinking Mode

Gemini 2.5 Pro / 3.1 Pro are thinking models that reason through their thoughts before responding.

4. Deep Research

Autonomously performs multi-source web research and generates comprehensive reports. Ideal for academic and business research.

5. Google Workspace Integration

Direct integration into Gmail, Docs, Sheets, and Slides. Seamless document generation, summarization, email drafting, and presentation creation.

6. Gemini Live

Real-time voice conversation mode with camera video sharing capability.

Context Length Pricing

⚠️ Gemini 2.5 Pro uses tiered pricing — when your total prompt exceeds 200K tokens, the entire request is billed at the higher rate, not just the tokens above the threshold.

Strengths

  • Native multimodal (video and audio included)
  • Google ecosystem integration (Search, Workspace, YouTube, Vertex AI)
  • Gemini 2.5 Flash's unbeatable low cost ($0.15/$0.60)
  • 1M token context (2M coming)
  • Enterprise-grade via Vertex AI
  • Gemini Live — real-time voice/video conversation

Weaknesses

  • ❌ Accuracy can be inconsistent
  • ❌ Frequent API changes
  • Tiered pricing above 200K tokens (more expensive for long contexts)
  • ❌ Regional/age restrictions

Official Resources

公式サイト