---
title: PDF Inputs
description: Send PDF documents to models through the Agnic AI Gateway
---

# PDF Inputs

Process PDF documents with compatible models through the Agnic AI Gateway. PDFs can be sent as direct URLs or base64-encoded data URLs.

## Overview

PDF processing is available via the `/v1/chat/completions` API using the `file` content type. This feature works with models that support file input.

<Callout type="info">
  When a model supports file input natively, the PDF is passed directly. Otherwise, the PDF is parsed and the text is passed to the model.
</Callout>

---

## Using PDF URLs

For publicly accessible PDFs, send the URL directly:

### Python

```python
from openai import OpenAI

client = OpenAI(
    api_key="agnic_tok_YOUR_TOKEN",
    base_url="https://api.agnic.ai/v1"
)

response = client.chat.completions.create(
    model="anthropic/claude-3.5-sonnet",
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "What are the main points in this document?"
                },
                {
                    "type": "file",
                    "file": {
                        "filename": "document.pdf",
                        "file_data": "https://example.com/document.pdf"
                    }
                }
            ]
        }
    ]
)

print(response.choices[0].message.content)
```

### JavaScript

```javascript
import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: 'agnic_tok_YOUR_TOKEN',
  baseURL: 'https://api.agnic.ai/v1'
});

const response = await client.chat.completions.create({
  model: 'anthropic/claude-3.5-sonnet',
  messages: [
    {
      role: 'user',
      content: [
        {
          type: 'text',
          text: 'What are the main points in this document?'
        },
        {
          type: 'file',
          file: {
            filename: 'document.pdf',
            file_data: 'https://example.com/document.pdf'
          }
        }
      ]
    }
  ]
});

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

### cURL

```bash
curl https://api.agnic.ai/v1/chat/completions \
  -H "Authorization: Bearer agnic_tok_YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-3.5-sonnet",
    "messages": [
      {
        "role": "user",
        "content": [
          {"type": "text", "text": "Summarize this document"},
          {
            "type": "file",
            "file": {
              "filename": "document.pdf",
              "file_data": "https://example.com/document.pdf"
            }
          }
        ]
      }
    ]
  }'
```

---

## Using Base64 Encoded PDFs

For local PDF files:

### Python

```python
from openai import OpenAI
import base64

def encode_pdf(pdf_path):
    with open(pdf_path, "rb") as pdf_file:
        return base64.b64encode(pdf_file.read()).decode('utf-8')

client = OpenAI(
    api_key="agnic_tok_YOUR_TOKEN",
    base_url="https://api.agnic.ai/v1"
)

# Read and encode the PDF
pdf_path = "path/to/your/document.pdf"
base64_pdf = encode_pdf(pdf_path)
data_url = f"data:application/pdf;base64,{base64_pdf}"

response = client.chat.completions.create(
    model="anthropic/claude-3.5-sonnet",
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "What are the key findings in this report?"
                },
                {
                    "type": "file",
                    "file": {
                        "filename": "report.pdf",
                        "file_data": data_url
                    }
                }
            ]
        }
    ]
)

print(response.choices[0].message.content)
```

### JavaScript

```javascript
import OpenAI from 'openai';
import fs from 'fs';

const client = new OpenAI({
  apiKey: 'agnic_tok_YOUR_TOKEN',
  baseURL: 'https://api.agnic.ai/v1'
});

// Read and encode the PDF
const pdfPath = 'path/to/your/document.pdf';
const pdfBuffer = fs.readFileSync(pdfPath);
const base64Pdf = pdfBuffer.toString('base64');
const dataUrl = `data:application/pdf;base64,${base64Pdf}`;

const response = await client.chat.completions.create({
  model: 'anthropic/claude-3.5-sonnet',
  messages: [
    {
      role: 'user',
      content: [
        {
          type: 'text',
          text: 'What are the key findings in this report?'
        },
        {
          type: 'file',
          file: {
            filename: 'report.pdf',
            file_data: dataUrl
          }
        }
      ]
    }
  ]
});

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

---

## Compatible Models

Models with PDF/file processing capabilities:

| Provider | Models | Notes |
|----------|--------|-------|
| Anthropic | Claude 3.5 Sonnet, Claude 3 Opus | Native PDF support |
| Google | Gemini 1.5 Pro, Gemini 2.0 Flash | Native file support |

---

## Use Cases

- **Document summarization** - Get key points from long documents
- **Data extraction** - Pull specific information from reports
- **Q&A over documents** - Ask questions about PDF content
- **Contract analysis** - Review legal documents
- **Research papers** - Analyze academic content

---

## Best Practices

1. **Use URLs when possible** - More efficient for large files
2. **Provide context** - Tell the model what to look for
3. **Break up large documents** - Split very long PDFs if needed
4. **Check model limits** - Different models have different page limits

<Callout type="tip">
  For scanned PDFs or documents with images, use models with strong OCR capabilities like Claude 3 or Gemini 1.5 Pro.
</Callout>

---

## Troubleshooting

**PDF not processing?**
- Verify the model supports file input
- Check that the PDF is not corrupted
- Ensure the file size is within limits

**Poor extraction results?**
- Try a model with better OCR capabilities
- Ensure the PDF has selectable text (not just images)

<Cards>
  <Card title="Image Inputs" href="/docs/ai-gateway/multimodal/images" />
  <Card title="Audio Inputs" href="/docs/ai-gateway/multimodal/audio" />
</Cards>