OpenRouter：连接300+模型的统一LLM API网关，节省40%成本 —— 2026年设置指南

# 安全存储你的API密钥
export OPENROUTER_API_KEY="sk-or-v1-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"

# 基本聊天补全请求
curl -s https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-sonnet-4.5",
    "messages": [
      {"role": "user", "content": "用3句话解释量子计算"}
    ]
  }'

# 无需特殊SDK —— 直接使用OpenAI客户端
pip install openai>=1.30.0

# openrouter_demo.py
from openai import OpenAI
import os

client = OpenAI(
    base_url="https://openrouter.ai/api/v1",
    api_key=os.environ.get("OPENROUTER_API_KEY"),
)

# 通过OpenRouter调用Claude Sonnet 4.5
response = client.chat.completions.create(
    model="anthropic/claude-sonnet-4.5",
    messages=[
        {"role": "system", "content": "你是一个有帮助的编程助手。"},
        {"role": "user", "content": "写一个Python函数来扁平化嵌套列表"}
    ],
    temperature=0.7,
    max_tokens=500,
)

print(response.choices[0].message.content)
print(f"使用的模型: {response.model}")
print(f"Token数: {response.usage.total_tokens}")

python openrouter_demo.py

npm install openai

// openrouter-demo.ts
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://openrouter.ai/api/v1",
  apiKey: process.env.OPENROUTER_API_KEY,
});

async function main() {
  const response = await client.chat.completions.create({
    model: "openai/gpt-5",
    messages: [
      { role: "user", content: "写一个React useDebounce hook" },
    ],
  });

  console.log(response.choices[0].message.content);
}

main();

# 列出所有300+模型及其定价
curl -s https://openrouter.ai/api/v1/models \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" | \
  jq '.data[] | {id: .id, pricing: .pricing}' | head -50

# openrouter_langchain.py
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate

# 创建指向OpenRouter的LangChain模型
llm = ChatOpenAI(
    model_name="anthropic/claude-sonnet-4.5",
    openai_api_key=os.environ.get("OPENROUTER_API_KEY"),
    openai_api_base="https://openrouter.ai/api/v1",
    temperature=0.7,
)

prompt = ChatPromptTemplate.from_messages([
    ("system", "你是Python开发专家。"),
    ("human", "{input}"),
])

chain = prompt | llm

result = chain.invoke({"input": "写一个FastAPI速率限制中间件"})
print(result.content)

# openrouter_llamaindex.py
from llama_index.llms.openai import OpenAI as LlamaOpenAI
from llama_index.core import Settings

llm = LlamaOpenAI(
    model="meta-llama/llama-4-maverick",
    api_key=os.environ.get("OPENROUTER_API_KEY"),
    api_base="https://openrouter.ai/api/v1",
    temperature=0.3,
)

Settings.llm = llm

# 正常使用LlamaIndex —— OpenRouter处理提供商连接
from llama_index.core import VectorStoreIndex, Document

docs = [Document(text="OpenRouter简化了多提供商LLM访问。")]
index = VectorStoreIndex.from_documents(docs)
query_engine = index.as_query_engine()
response = query_engine.query("OpenRouter是做什么的？")
print(response)

// app/api/chat/route.ts
import { createOpenRouter } from "@openrouter/ai-sdk-provider";
import { convertToModelMessages, streamText } from "ai";

const openrouter = createOpenRouter({
  apiKey: process.env.OPENROUTER_API_KEY,
});

export async function POST(req: Request) {
  const { messages } = await req.json();

  const result = streamText({
    model: openrouter("anthropic/claude-sonnet-4.5"),
    messages: await convertToModelMessages(messages),
    system: "你是一个有用的助手。",
  });

  return result.toDataStreamResponse();
}

// openrouter_demo.go
package main

import (
	"context"
	"fmt"
	"os"

	"github.com/openai/openai-go"
	"github.com/openai/openai-go/option"
)

func main() {
	client := openai.NewClient(
		option.WithBaseURL("https://openrouter.ai/api/v1"),
		option.WithAPIKey(os.Getenv("OPENROUTER_API_KEY")),
	)

	resp, err := client.Chat.Completions.New(context.Background(), openai.ChatCompletionNewParams{
		Model: openai.String("google/gemini-3-pro"),
		Messages: openai.F([]openai.ChatCompletionMessageParamUnion{
			openai.UserMessage("解释Go并发模式"),
		}),
	})
	if err != nil {
		panic(err)
	}

	fmt.Println(resp.Choices[0].Message.Content)
}

# 让OpenRouter自动选择最佳模型
response = client.chat.completions.create(
    model="openrouter/auto",  # 从58+候选模型中自动选择
    messages=[
        {"role": "user", "content": "写一个Kubernetes部署YAML"}
    ],
    # 可选：添加路由偏好
    extra_body={
        "provider": {
            "sort": "price",  # 或 "throughput", "latency"
        }
    }
)
print(response.model)  # 显示实际使用了哪个模型

# 生产故障转移配置
response = client.chat.completions.create(
    model="anthropic/claude-sonnet-4.5",
    messages=[{"role": "user", "content": "关键业务分析..."}],
    extra_body={
        "provider": {
            "order": ["Anthropic", "OpenAI", "Google"],
            "allow_fallbacks": True,
        },
        "models": [
            "anthropic/claude-sonnet-4.5",
            "openai/gpt-5",
            "google/gemini-3-pro",
        ]
    }
)

# 存储你的直接提供商密钥
curl -X POST https://openrouter.ai/api/v1/credentials \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "provider": "openai",
    "key": "sk-proj-your-direct-openai-key"
  }'

# 路由到最便宜的可用模型
response = client.chat.completions.create(
    model="openrouter/auto",
    messages=[{"role": "user", "content": "总结这篇文章"}],
    extra_body={
        "provider": {
            "sort": "price",
            "quantizations": ["fp8", "fp16"],  # 优先量化模型
        }
    }
)

# 路由到最快的模型
response = client.chat.completions.create(
    model="openrouter/auto",
    messages=[{"role": "user", "content": "快速是/否问题"}],
    extra_body={
        "provider": {
            "sort": "throughput",
        }
    }
)

# Dockerfile.openrouter-proxy
FROM node:20-alpine

WORKDIR /app
COPY package*.json ./
RUN npm install express axios

COPY . .
EXPOSE 3000
CMD ["node", "proxy.js"]

# docker-compose.yml
version: "3.8"
services:
  openrouter-proxy:
    build:
      context: .
      dockerfile: Dockerfile.openrouter-proxy
    ports:
      - "3000:3000"
    environment:
      - OPENROUTER_API_KEY=${OPENROUTER_API_KEY}
      - FALLBACK_MODELS=openai/gpt-5,google/gemini-3-pro
      - CACHE_ENABLED=true
    restart: unless-stopped

# 以编程方式跟踪使用和成本
import requests

headers = {"Authorization": f"Bearer {os.environ.get('OPENROUTER_API_KEY')}"}

# 获取使用统计
usage = requests.get(
    "https://openrouter.ai/api/v1/credits",
    headers=headers
).json()

print(f"剩余信用额: ${usage['data']['total_credits'] - usage['data']['total_usage']}")
print(f"总使用量: ${usage['data']['total_usage']}")

# 相同代码，不同模型
model = "anthropic/claude-sonnet-4.5"  # 或 "openai/gpt-5" 或 "google/gemini-3-pro"
response = client.chat.completions.create(
    model=model,
    messages=[{"role": "user", "content": "你好！"}]
)