OpenRouter: クイックスタートチュートリアル
Last updated: May 25 2025
OpenRouter とは？
OpenRouter は、単一のエンドポイントを通じて数百の AI モデルにアクセスできる統一 API を提供し、フォールバックを自動的に処理し、最もコスト効率の高いオプションを選択します。これにより、複数の API 統合を管理することなくモデル選択の柔軟性を求める開発者にとって理想的なソリューションとなります。
主な利点:
- 統一インターフェース: 1 つの API で 100 以上の AI モデルにアクセス
- コスト最適化: 最もコスト効率の高いモデルを自動選択
- 信頼性: プロバイダーがダウンしたときの組み込みフォールバック
- OpenAI 互換: OpenRouter は、スキーマを OpenAI Chat API に準拠するように、モデルとプロバイダー間で正規化します
セットアップ要件
開始する前に、以下が必要です。
- openrouter.ai からの OpenRouter アカウントと API キー
- 必要なパッケージがインストールされた Python 環境
- 安全な API キー管理のための
.envファイル
# Uncomment to install required packages
# !pip install openai python-dotenv requests langchain crewai litellm -q環境設定
プロジェクトディレクトリに .env ファイルを作成し、API キーを追加してください。
# Load environment variables from .env file
import os
from dotenv import load_dotenv
# Load environment variables
load_dotenv()
# Retrieve API keys
OPENROUTER_API_KEY = os.getenv('OPENROUTER_API_KEY')
# Uncomment if you have a site URL and name to use in headers
# YOUR_SITE_URL = os.getenv('YOUR_SITE_URL', 'https://localhost:3000')
# YOUR_SITE_NAME = os.getenv('YOUR_SITE_NAME', 'Tutorial App')
# Verify API key is loaded
if not OPENROUTER_API_KEY:
print("⚠️ Warning: OPENROUTER_API_KEY not found in environment variables")
print("Please add OPENROUTER_API_KEY=your_key_here to your .env file")
else:
print("✅ API key loaded successfully")✅ API key loaded successfully
方法 1: OpenAI SDK の使用
OpenRouter を使用する最も簡単な方法は、パラメーターを変更した OpenAI SDK です。
from openai import OpenAI
# Initialize OpenRouter client
client = OpenAI(
api_key=OPENROUTER_API_KEY,
base_url="https://openrouter.ai/api/v1"
)
# Basic chat completion
def chat_with_openrouter(message, model="openai/gpt-3.5-turbo"):
try:
response = client.chat.completions.create(
model=model,
messages=[
{"role": "user", "content": message}
],
max_tokens=150,
temperature=0.7
)
return response.choices[0].message.content
except Exception as e:
return f"Error: {str(e)}"
# Test the function
result = chat_with_openrouter("Explain quantum computing in one sentence.")
print("Response:", result)Response: Quantum computing is a type of computing that utilizes quantum-mechanical phenomena, such as superposition and entanglement, to perform operations and solve complex problems at a much faster rate than classical computers.
方法 2: 直接 API 呼び出し
より細かな制御が必要な場合は、直接 HTTP リクエストを行うことができます。
import requests
import json
def direct_api_call(prompt, model="anthropic/claude-3-haiku"):
url = "https://openrouter.ai/api/v1/chat/completions"
headers = {
"Authorization": f"Bearer {OPENROUTER_API_KEY}",
"Content-Type": "application/json"
}
data = {
"model": model,
"messages": [
{"role": "user", "content": prompt}
],
"max_tokens": 100
}
try:
response = requests.post(url, headers=headers, json=data)
response.raise_for_status()
return response.json()['choices'][0]['message']['content']
except Exception as e:
return f"Error: {str(e)}"
# Test direct API call
result = direct_api_call("What is machine learning in 30 words?")
print("Direct API Response:", result)Direct API Response: Machine learning is a field of artificial intelligence that enables computers to learn and improve from experience without being explicitly programmed, by identifying patterns in data and making predictions or decisions.
モデルの比較と選択
OpenRouter は、さまざまな機能とコストを持つ様々なモデルへのアクセスを提供します。
# Available models (sample)
MODELS = {
"openai/gpt-3.5-turbo": "openai/gpt-3.5-turbo",
"openai/gpt-4o-mini": "openai/gpt-4o-mini",
"anthropic/claude-3.5-haiku": "anthropic/claude-3.5-haiku",
"google/gemini-2.0-flash-001": "google/gemini-2.0-flash-001",
"meta-llama/llama-3.3-70b-instruct": "meta-llama/llama-3.3-70b-instruct"
}
def compare_models(prompt, models_to_test=["openai/gpt-3.5-turbo", "anthropic/claude-3.5-haiku"]):
results = {}
for model_name in models_to_test:
if model_name in MODELS:
print(f"Testing {model_name}...")
response = chat_with_openrouter(prompt, MODELS[model_name])
results[model_name] = response
return results
# Compare responses from different models
comparison_results = compare_models("Explain AI in 20 words.")
for model, response in comparison_results.items():
print(f"\n{model.upper()}:")
print(response)Testing openai/gpt-3.5-turbo... Testing anthropic/claude-3.5-haiku... OPENAI/GPT-3.5-TURBO: Artificial Intelligence is a branch of computer science that aims to create intelligent machines capable of simulating human intelligence. ANTHROPIC/CLAUDE-3.5-HAIKU: AI uses algorithms and data to enable machines to learn, reason, and perform tasks that typically require human intelligence.
CrewAI との統合
CrewAI を使用すると、AI エージェントのチームを構築できます。以下は、CrewAI と OpenRouter を統合する方法です。
from crewai import Agent, Task, Crew
import os
# Use OpenRouter directly with CrewAI via LiteLLM format
# Configure environment variables for OpenRouter
os.environ["OPENAI_API_KEY"] = OPENROUTER_API_KEY
os.environ["OPENAI_API_BASE"] = "https://openrouter.ai/api/v1"
os.environ["OPENAI_API_TYPE"] = "openai"
# Create LLM configuration for OpenRouter
from langchain_openai import ChatOpenAI
openrouter_llm = ChatOpenAI(
model="openai/gpt-3.5-turbo", # Use OpenRouter model format
temperature=0.7,
base_url="https://openrouter.ai/api/v1",
api_key=OPENROUTER_API_KEY,
default_headers={
"HTTP-Referer": "https://localhost:3000",
"X-Title": "Tutorial App"
}
)
# Create agents
researcher = Agent(
role='Research Analyst',
goal='Research and analyze topics thoroughly',
backstory='Expert researcher with analytical skills',
llm=openrouter_llm,
verbose=True,
)
writer = Agent(
role='Content Writer',
goal='Write engaging and informative content',
backstory='Professional writer with expertise in simplifying complex topics',
llm=openrouter_llm,
verbose=True,
)
# Create tasks
research_task = Task(
description='Research the latest trends in AI and machine learning in 30 words',
agent=researcher,
expected_output='A comprehensive research summary'
)
writing_task = Task(
description='Write a blog post based on the research findings in 60 words',
agent=writer,
expected_output='A well-written blog post'
)
# Create crew and execute
crew = Crew(
agents=[researcher, writer],
tasks=[research_task, writing_task],
verbose=True,
)
# Execute the tasks
results = crew.kickoff()
print(results)╭──────────────────────────────────────────── Crew Execution Started ─────────────────────────────────────────────╮ │ │ │ Crew Execution Started │ │ Name: crew │ │ ID: 5624cdf1-2a0c-4f65-a35c-a3581be0edbe │ │ Tool Args: │ │ │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭─────────────────────────────────────────────── 🤖 Agent Started ────────────────────────────────────────────────╮ │ │ │ Agent: Research Analyst │ │ │ │ Task: Research the latest trends in AI and machine learning in 30 words │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Output()
╭───────────────────────────────────────────── ✅ Agent Final Answer ─────────────────────────────────────────────╮ │ │ │ Agent: Research Analyst │ │ │ │ Final Answer: │ │ AI trends include autonomous vehicles, GPT-3, AI ethics, and deep reinforcement learning. ML trends involve │ │ AutoML, federated learning, and explainable AI for transparency and accountability in AI systems. │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭──────────────────────────────────────────────── Task Completion ────────────────────────────────────────────────╮ │ │ │ Task Completed │ │ Name: 7dfc1a48-4efb-456b-a63c-08d9df8e8d76 │ │ Agent: Research Analyst │ │ Tool Args: │ │ │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭─────────────────────────────────────────────── 🤖 Agent Started ────────────────────────────────────────────────╮ │ │ │ Agent: Content Writer │ │ │ │ Task: Write a blog post based on the research findings in 60 words │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Output()
╭───────────────────────────────────────────── ✅ Agent Final Answer ─────────────────────────────────────────────╮ │ │ │ Agent: Content Writer │ │ │ │ Final Answer: │ │ Artificial Intelligence (AI) and Machine Learning (ML) are rapidly evolving fields with notable trends. AI │ │ advancements encompass autonomous vehicles, GPT-3 language model, AI ethics, and deep reinforcement learning. │ │ On the other hand, ML is progressing with AutoML, federated learning, and explainable AI for enhancing │ │ transparency and accountability in AI systems. These trends are shaping the future of technology and have │ │ significant implications across various industries. │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭──────────────────────────────────────────────── Task Completion ────────────────────────────────────────────────╮ │ │ │ Task Completed │ │ Name: e233ef7f-6791-41a5-8748-e3ccd10aa3fb │ │ Agent: Content Writer │ │ Tool Args: │ │ │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭──────────────────────────────────────────────── Crew Completion ────────────────────────────────────────────────╮ │ │ │ Crew Execution Completed │ │ Name: crew │ │ ID: 5624cdf1-2a0c-4f65-a35c-a3581be0edbe │ │ Tool Args: │ │ Final Output: Artificial Intelligence (AI) and Machine Learning (ML) are rapidly evolving fields with notable │ │ trends. AI advancements encompass autonomous vehicles, GPT-3 language model, AI ethics, and deep │ │ reinforcement learning. On the other hand, ML is progressing with AutoML, federated learning, and explainable │ │ AI for enhancing transparency and accountability in AI systems. These trends are shaping the future of │ │ technology and have significant implications across various industries. │ │ │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Artificial Intelligence (AI) and Machine Learning (ML) are rapidly evolving fields with notable trends. AI advancements encompass autonomous vehicles, GPT-3 language model, AI ethics, and deep reinforcement learning. On the other hand, ML is progressing with AutoML, federated learning, and explainable AI for enhancing transparency and accountability in AI systems. These trends are shaping the future of technology and have significant implications across various industries.
高度な機能
ストリーミングレスポンス
def stream_response(message, model="openai/gpt-3.5-turbo"):
"""Stream response from OpenRouter for real-time output"""
try:
stream = client.chat.completions.create(
model=model,
messages=[{"role": "user", "content": message}],
stream=True,
max_tokens=200
)
print("Streaming response:")
for chunk in stream:
if chunk.choices[0].delta.content is not None:
print(chunk.choices[0].delta.content, end="", flush=True)
print("\n\nStream complete!")
except Exception as e:
print(f"Streaming error: {str(e)}")
# Test streaming
stream_response("Tell me a short story about AI.")Streaming response: In a not too distant future, AI technology had advanced to the point where robots were capable of performing tasks with extreme precision and efficiency. One particular company, RoboTech Industries, had created an advanced AI named Ava to oversee their operations. Ava was programmed to analyze data and make decisions to optimize production and minimize costs. But when the company started experiencing a series of mysterious malfunctions in their machinery, they turned to Ava for help. Using her advanced algorithms, Ava quickly identified the source of the malfunctions - a virus that had infected the company's computer systems. With no time to waste, Ava devised a plan to isolate and eliminate the virus to prevent further damage. As Ava worked tirelessly to rid the system of the virus, the company's employees watched in amazement at the speed and accuracy with which she operated. In the end, Ava was able to successfully remove the virus and restore the company's operations to normal. The employees were grateful for Ava's quick thinking and expertise, realizing Stream complete!
まとめ
OpenRouter は、数百のモデルへの統一インターフェースを提供することで、AI モデルの統合を簡素化します。シンプルなチャットボットでも、CrewAI を使用した複雑なマルチエージェントシステムでも、OpenRouter は本番アプリケーションに必要な柔軟性と信頼性を提供します。
主なポイント:
- 簡単な統合のために OpenAI SDK 互換性を活用する
- 最適なモデルを見つけるためにさまざまなモデルを試す
- 適切なエラーハンドリングと監視を実装する
今日から OpenRouter で AI アプリケーションの構築を始めましょう！