API_USAGE.md

LiteLLM API 使用指南

基础信息

配置项	值
API Base URL	https://litellm.example.com
认证方式	Bearer Token (LiteLLM Native Auth)
API Key 存储	RDS PostgreSQL
Key 创建方式	LiteLLM Admin UI (/ui)

---

可用模型

模型名称	端点	用途
claude-opus-4-6-us	us	最强，us 端点
claude-opus-4-6-global	global	最强，global 端点
claude-opus-4-6	us	Opus 4.6 兼容别名
claude-opus-4-5	global	次强
claude-sonnet-4-6-us	us	最新 Sonnet，us 端点
claude-sonnet-4-6-global	global	最新 Sonnet，global 端点
claude-sonnet-4-6	us	Sonnet 4.6 兼容别名
claude-sonnet-4-5	global	平衡性能与成本
claude-sonnet-3-7	us	性价比高
claude-sonnet-3-7	us	经典可靠
claude-haiku-4-5	global	最快最便宜

---

1. cURL 调用

Chat Completions

curl -X POST https://litellm.example.com/v1/chat/completions \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-5",
    "messages": [{"role": "user", "content": "Hello"}],
    "max_tokens": 1000
  }'

注意: <YOUR_API_KEY> 是通过 LiteLLM Admin UI (/ui) 创建的 API Key (sk-xxx 格式)。

Streaming

curl -X POST https://litellm.example.com/v1/chat/completions \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-5",
    "messages": [{"role": "user", "content": "Hello"}],
    "max_tokens": 1000,
    "stream": true
  }'

列出模型

curl https://litellm.example.com/v1/models \
  -H "Authorization: Bearer <YOUR_API_KEY>"

---

2. Python (OpenAI SDK)

安装

pip install openai

基本调用

from openai import OpenAI

client = OpenAI(
    base_url="https://litellm.example.com/v1",
    api_key="<YOUR_API_KEY>"
)

response = client.chat.completions.create(
    model="claude-sonnet-4-5",
    messages=[{"role": "user", "content": "Hello"}],
    max_tokens=1000
)
print(response.choices[0].message.content)

Streaming

from openai import OpenAI

client = OpenAI(
    base_url="https://litellm.example.com/v1",
    api_key="<YOUR_API_KEY>"
)

stream = client.chat.completions.create(
    model="claude-sonnet-4-5",
    messages=[{"role": "user", "content": "Write a poem"}],
    max_tokens=500,
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

多轮对话

from openai import OpenAI

client = OpenAI(
    base_url="https://litellm.example.com/v1",
    api_key="<YOUR_API_KEY>"
)

messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "What is Python?"}
]

response = client.chat.completions.create(
    model="claude-sonnet-4-5",
    messages=messages,
    max_tokens=1000
)

# 添加助手回复到历史
messages.append({"role": "assistant", "content": response.choices[0].message.content})

# 继续对话
messages.append({"role": "user", "content": "Show me an example"})

response = client.chat.completions.create(
    model="claude-sonnet-4-5",
    messages=messages,
    max_tokens=1000
)
print(response.choices[0].message.content)

---

3. Python (Anthropic SDK)

安装

pip install anthropic

基本调用

import anthropic

client = anthropic.Anthropic(
    base_url="https://litellm.example.com",
    api_key="<YOUR_API_KEY>"
)

message = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=1000,
    messages=[{"role": "user", "content": "Hello"}]
)
print(message.content[0].text)

Streaming

import anthropic

client = anthropic.Anthropic(
    base_url="https://litellm.example.com",
    api_key="<YOUR_API_KEY>"
)

with client.messages.stream(
    model="claude-sonnet-4-5",
    max_tokens=1000,
    messages=[{"role": "user", "content": "Hello"}]
) as stream:
    for text in stream.text_stream:
        print(text, end="")

---

4. Node.js (OpenAI SDK)

安装

npm install openai

基本调用

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://litellm.example.com/v1',
  apiKey: '<YOUR_API_KEY>'
});

async function main() {
  const response = await client.chat.completions.create({
    model: 'claude-sonnet-4-5',
    messages: [{ role: 'user', content: 'Hello' }],
    max_tokens: 1000
  });
  console.log(response.choices[0].message.content);
}

main();

Streaming

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://litellm.example.com/v1',
  apiKey: '<YOUR_API_KEY>'
});

async function main() {
  const stream = await client.chat.completions.create({
    model: 'claude-sonnet-4-5',
    messages: [{ role: 'user', content: 'Hello' }],
    max_tokens: 1000,
    stream: true
  });

  for await (const chunk of stream) {
    process.stdout.write(chunk.choices[0]?.delta?.content || '');
  }
}

main();

---

5. 环境变量配置

OpenAI SDK 兼容

export OPENAI_API_BASE="https://litellm.example.com/v1"
export OPENAI_API_KEY="<YOUR_API_KEY>"

Anthropic SDK 兼容

export ANTHROPIC_BASE_URL="https://litellm.example.com/"
export ANTHROPIC_API_KEY="<YOUR_API_KEY>"

Claude Code 配置

~/.claude/settings.json:

{
  "apiKeyHelper": "echo <YOUR_API_KEY>",
  "primaryProvider": "anthropic"
}

环境变量:

export ANTHROPIC_BASE_URL="https://litellm.example.com/"

---

6. 健康检查

# Liveness
curl https://litellm.example.com/health/liveliness

# Readiness
curl https://litellm.example.com/health/readiness

---

7. Fallback 降级策略

当主模型不可用时，自动降级到备选模型：

• claude-opus-4-6-us -> claude-opus-4-6-global -> claude-opus-4-5 -> claude-sonnet-4-6-us -> claude-sonnet-4-6-global -> claude-sonnet-4-5
• claude-sonnet-4-5 -> claude-sonnet-3-7 -> claude-sonnet-4-6-us
• claude-haiku-4-5 -> claude-sonnet-4-6-us