# Anthropic LangChain Integration

> Use LangChain to integrate Anthropic with Helicone to log your Anthropic LLM usage.

import LegacyWarning from "/snippets/legacy-provider-warning.mdx";

<LegacyWarning />

<Steps>
  <Step title="Create an account + Generate an API Key">
    Log into [helicone](https://www.helicone.ai) or create an account. Once you have an account, you
    can generate an [API key](https://helicone.ai/developer).
  </Step>
  <Step title="Set HELICONE_API_KEY as an environment variable">
```typescript
export HELICONE_API_KEY=<your API key>
```
  </Step>
  <Step title="Modify the API base and add the `Helicone-Auth` header">
<CodeGroup>

```typescript example.ts
const llm = new ChatAnthropic({
  modelName: "claude-2",
  anthropicApiKey: "ANTHROPIC_API_KEY",
  clientOptions: {
    baseURL: "https://anthropic.helicone.ai",
    defaultHeaders: {
      "Helicone-Auth": `Bearer ${HELICONE_API_KEY}`,
    },
  },
});
```

```python example.py
anthropic = ChatAnthropic(
  temperature=0.9,
  model="claude-2",
  anthropic_api_url="https://anthropic.helicone.ai",
  anthropic_api_key="ANTHROPIC_API_KEY",
  model_kwargs={
    "extra_headers":{
      "Helicone-Auth": f"Bearer {HELICONE_API_KEY}"
    }
  }
)
```

</CodeGroup>

  </Step>
</Steps>

## Using AnthropicMultiModal with Helicone

Enhance your integration by using **AnthropicMultiModal** with Helicone to process images and generate text descriptions. Follow the code below to add this functionality to your application.

```python
import os
from llama_index.multi_modal_llms.anthropic import AnthropicMultiModal
from llama_index.core.multi_modal_llms.generic_utils import load_image_urls

# Initialize the AnthropicMultiModal class with your Anthropic API key
anthropic_mm_llm = AnthropicMultiModal(
    max_tokens=300,
    default_headers={
        "Helicone-Auth": f"Bearer {os.environ['HELICONE_API_KEY']}",
    },
    api_key="<your Anthropic API key>",
    api_base="https://anthropic.helicone.ai",
)

# Provide image URLs
image_urls = ["https://example.com/image1.png"]

# Load image URLs into documents
image_url_documents = load_image_urls(image_urls)

# Generate a response based on the images
response = anthropic_mm_llm.complete(
    prompt="Describe the images as alternative text",
    image_documents=image_url_documents,
)

print(response)
```