Skip to main content

​
Key Highlights

  • World’s Best: LMArena leaderboard 1501 Elo, ranked #1 globally
  • Top Coding: 76.2% on SWE-bench Verified, industry-leading code generation
  • Massive Context: 1 million token context window for large codebases
  • Chain-of-Thought: Display complete reasoning process
  • Great Value: Official pricing + up to 20% discount with recharge bonuses

​
Background

On November 18, 2025, Google officially released Gemini 3 Pro Preview, their latest multimodal AI model. With a score of 1501 Elo on the LMArena leaderboard, it ranks #1 globally, surpassing GPT-5, Claude Opus 4.1, and all other competitors. API易 has integrated this model immediately, providing developers with stable, cost-effective API services.

​
Key Features

​
New Models

gemini-3-pro-preview

Auto Reasoning ModelAutomatically adjusts reasoning intensity based on task complexity.

gemini-3-pro-preview-thinking

Thinking Output ModelDisplays complete reasoning process, ideal for advanced programming and mathematics.

​
Performance Benchmarks

BenchmarkScoreRank
LMArena1501 Elo🏆 #1 Global
SWE-bench Verified76.2%🥇 Top Tier
Terminal-Bench 2.054.2%🥇 Leading
Multimodal-🏆 Best in Class

​
Core Capabilities

  • 1M Token Context: Handle massive codebases and documents
  • 64K Token Output: Generate extensive code and documentation
  • Multimodal: Text, image, video, and audio understanding
  • Tool Integration: Google Search, File Search, Code Execution, Function Calling

​
Use Cases

AI Programming

  • IDE integration
  • Code review
  • Autonomous development
  • Technical documentation

Complex Reasoning

  • Mathematical proofs
  • Logic reasoning
  • Strategy planning
  • Data analysis

Multimodal Apps

  • Image understanding
  • Video analysis
  • OCR and parsing
  • Cross-modal generation

Agent Development

  • Agentic workflows
  • Autonomous tasks
  • Tool integration
  • Context management

​
Code Example

import openai

client = openai.OpenAI(
    api_key="your-api-key",
    base_url="https://api.apiyi.com/v1"
)

# Basic chat
response = client.chat.completions.create(
    model="gemini-3-pro-preview",
    messages=[
        {"role": "user", "content": "Explain quantum entanglement"}
    ]
)
print(response.choices[0].message.content)

# Chain-of-thought mode
response = client.chat.completions.create(
    model="gemini-3-pro-preview-thinking",
    messages=[
        {"role": "user", "content": "Prove: 1+2+3+...+n = n(n+1)/2"}
    ]
)
print(response.choices[0].message.content)

​
Pricing

ItemPriceNotes
Input Tokens$2.00 / 1M tokensFor prompts ≤200K tokens
Output Tokens$12.00 / 1M tokensAll output tokens
Cached Input$0.20 / 1M tokensIf caching supported
Special Offer: With API易’s recharge bonus program, enjoy up to 20% discount! Input costs as low as $1.6/1M tokens, output costs as low as $9.6/1M tokens.

​
Getting Started

  1. Sign up: https://api.apiyi.com
  2. Add credits: Enjoy recharge bonuses
  3. Check docs: Gemini API Guide
  4. Start using: Change model to gemini-3-pro-preview
Sources:
  • Google Official Blog: Gemini 3 Pro Preview Announcement blog.google/technology/ai/google-gemini-3-pro-preview/
  • LMArena Leaderboard: lmarena.ai/leaderboard
  • SWE-bench: swebench.com
  • Data retrieved: November 20, 2025