Claude 兼容接入

Claude-Compatible Proxy

很多开发者不是不会调用 Claude，而是卡在另一层问题上：Claude 怎么接进现有项目、能不能保留现在的 OpenAI SDK、Claude Code 怎么和业务接口统一起来。 Most developers know how to call Claude. The harder question is: how do you integrate it into an existing project without rewriting your SDK layer, and how do you unify Claude Code with your backend calls.

如果你希望直接把 Claude 放进统一入口里，可以使用 api.llapi.net。一套 Key 加一条兼容地址，就能把 Claude、Claude Code 以及后续的 GPT、Gemini 一起收进同一套工作流。 To bring Claude into a unified gateway, use api.llapi.net. One key, one compatible endpoint — Claude, Claude Code, GPT, and Gemini all in the same workflow.

TL;DR

• 核心不是"替代 Claude"，而是把 Claude 收进统一调用入口
• The goal isn't to replace Claude — it's to route it through a unified entry point
• OpenAI SDK 项目：只改 apiKey、baseURL、model
• OpenAI SDK projects: only change apiKey, baseURL, model
• Claude Code：设置 ANTHROPIC_BASE_URL 和 ANTHROPIC_API_KEY 即可
• Claude Code: set ANTHROPIC_BASE_URL and ANTHROPIC_API_KEY

用 OpenAI SDK 调 ClaudeCalling Claude via OpenAI SDK

如果你的项目已经在用 OpenAI SDK，这是改造成本最低的方式。只改三处：

If your project already uses the OpenAI SDK, this is the lowest-cost path. Change only three things:

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.LLAPI_API_KEY,    // ← 换成灵络API KeyLingluoAPI key
  baseURL: "https://api.llapi.net/v1"     // ← 换成灵络API endpointLingluoAPI endpoint
});

const completion = await client.chat.completions.create({
  model: "claude-sonnet",               // ← 换成 Claude 模型名Claude model name
  messages: [
    { role: "system", content: "You are a concise engineering assistant." },
    { role: "user", content: "Explain why a proxy layer reduces migration cost." }
  ]
});

console.log(completion.choices[0]?.message?.content);

业务层的消息结构和调用方式完全不变，只有入口和模型名需要更新。

Your business-layer message structure and call patterns stay identical. Only the entry point and model name update.

用原生 Anthropic SDKNative Anthropic SDK

如果你想使用 Anthropic 原生 SDK 并直接调用 Claude API 格式，只需替换 base_url：

If you prefer the native Anthropic SDK and want to call Claude's API format directly, just swap the base_url:

import anthropic

client = anthropic.Anthropic(
    api_key="your-llapi-key",
    base_url="https://api.llapi.net"    # 注意：不带 /v1No /v1 suffix here
)

message = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello, Claude"}]
)
print(message.content)

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.LLAPI_API_KEY,
  baseURL: "https://api.llapi.net"
});

const msg = await client.messages.create({
  model: "claude-sonnet-4-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello, Claude" }]
});

console.log(msg.content);

Claude Code

如果你在用 Claude Code，只需设置两个环境变量，就能把 Claude Code 的请求路由到灵络API，与其他业务调用统一管理：

Using Claude Code? Set two environment variables to route all Claude Code requests through LingluoAPI, unifying it with your other API calls:

export ANTHROPIC_API_KEY="your-llapi-key"
export ANTHROPIC_BASE_URL="https://api.llapi.net"

# 永久生效（写入 shell 配置）Permanent (write to shell config)
echo 'export ANTHROPIC_API_KEY="your-llapi-key"' >> ~/.zshrc
echo 'export ANTHROPIC_BASE_URL="https://api.llapi.net"' >> ~/.zshrc
source ~/.zshrc

Python 流式输出Python Streaming

流式输出同样支持，写法与官方 SDK 一致：

Streaming is supported. The syntax is identical to the official SDK:

import anthropic

client = anthropic.Anthropic(
    api_key="your-llapi-key",
    base_url="https://api.llapi.net"
)

with client.messages.stream(
    model="claude-sonnet-4-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Write a haiku"}]
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)

适合哪些团队Who this is for

FAQ