litellm

通过 LiteLLM 的统一 API 呼叫 100 多家 LLM 提供商。当您需要调用与主模型不同的模型（例如，在 Claude 上运行时使用 GPT-4 进行代码审查）、比较多个模型的输出、路由到更便宜的模型来执行简单任务或访问运行时本身不支持的模型时使用。

## 概述（中文）

通过 LiteLLM 的统一 API 呼叫 100 多家 LLM 提供商。当您需要调用与主模型不同的模型（例如，在 Claude 上运行时使用 GPT-4 进行代码审查）、比较多个模型的输出、路由到更便宜的模型来执行简单任务或访问运行时本身不支持的模型时使用。

## 原文

# LiteLLM - Multi-Model LLM Calls

Use LiteLLM when you need to call LLMs beyond your primary model.

## When to Use

- **Model comparison**: Get outputs from multiple models and compare
- **Specialized routing**: Use code-optimized models for code, writing models for prose
- **Cost optimization**: Route simple queries to cheaper models
- **Fallback access**: Access models your runtime doesn't support

## Quick Start

```python
import litellm

# Call any model with unified API
response = litellm.completion(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Explain this code"}]
)
print(response.choices[0].message.content)
```

## Common Patterns

### Compare Multiple Models

```python
import litellm

prompt = [{"role": "user", "content": "What's the best approach to X?"}]

models = ["gpt-4o", "claude-sonnet-4-20250514", "gemini/gemini-1.5-pro"]
for model in models:
    resp = litellm.completion(model=model, messages=prompt)
    print(f"{model}: {resp.choices[0].message.content[:200]}...")
```

### Route by Task Type

```python
import litellm

def smart_call(task_type: str, prompt: str) -> str:
    model_map = {
        "code": "gpt-4o",           # Strong at code
        "writing": "claude-sonnet-4-20250514",  # Strong at prose
        "simple": "gpt-4o-mini",    # Cheap for simple tasks
        "reasoning": "o1-preview",  # Deep reasoning
    }
    model = model_map.get(task_type, "gpt-4o")
    resp = litellm.completion(
        model=model,
        messages=[{"role": "user", "content": prompt}]
    )
    return resp.choices[0].message.content
```

### Use LiteLLM Proxy (Recommended)

If a LiteLLM proxy is available, point to it for caching, rate limiting, and observability:

```python
import litellm

litellm.api_base = "https://your-litellm-proxy.com"
litellm.api_key = "sk-your-key"

response = litellm.completion(
    model="gpt-4o",  # Proxy routes to configured provider
    messages=[{"role": "user", "content": "Hello"}]
)
```

## Environment Setup

Ensure `litellm` is installed and API keys are set:

```bash
pip install litellm

# Set provider keys (or configure in proxy)
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY="sk-..."
```

## Model Reference

Common model identifiers:
- **OpenAI**: `gpt-4o`, `gpt-4o-mini`, `o1-preview`, `o1-mini`
- **Anthropic**: `claude-sonnet-4-20250514`, `claude-opus-4-20250514`
- **Google**: `gemini/gemini-1.5-pro`, `gemini/gemini-1.5-flash`
- **Mistral**: `mistral/mistral-large-latest`

Full list: https://docs.litellm.ai/docs/providers