Claude (OpenAI format) — PDF Analysis

Short description: Claude supports OpenAI-style requests while remaining compatible with the native Claude format. This page demonstrates how to upload PDFs or images using OpenAI-style requests for analysis. For caching or more efficient PDF handling, consider using the native Claude format.

---

Overview

• Method: POST
• Path: /v1/chat/completions
• Content-Type: application/json

Authentication

Supports common header-based authentication:

• OpenAI: Authorization: Bearer sk-xxx
• Anthropic (native): x-api-key: sk-xxx (native format may have additional fields)

---

Features

• Supports image and PDF analysis: include text and file items in messages. File items can be base64 (data:) or a file/URL reference.
• Recommendation: if you need to reuse the same PDF in context multiple times, use the native Claude format to leverage caching and faster parsing.

---

Request Examples

Body parameters

Parameter	Type	Required	Description
model	string	yes	Model name, e.g. claude-3-5-sonnet-20240620
messages	<object>	yes	Chat message array; items may be plain text or mixed arrays (text/file/image)
temperature	number	no	Sampling temperature (0-2)
top_p	number	no	Nucleus sampling threshold
stream	boolean	no	Whether to stream responses
max_tokens	number	no	Max generated tokens
thinking	object	no	Claude-specific "think first" parameter (supported by some models)

About possible content items in messages:

• Text: { "type": "text", "text": "..." } or plain string.
• File (base64 or URL):
  • Format 1 (file): { "type": "file", "file": { "filename": "api-doc.pdf", "file_data": "<base64 or URL>" } }
  • Format 2 (file_url): { "type": "file_url", "file_url": { "url": "https://..." } }
• Image: { "type": "image_url", "image_url": { "url": "https://..." } }

Example (chat):

{
  "model": "claude-3-5-sonnet-20240620",
  "messages": [{ "role": "user", "content": "Hello, who are you?" }],
  "max_tokens": 1688,
  "temperature": 0.5,
  "stream": false
}

Example (PDF analysis, format 1: file_data can be URL or base64):

{
  "model": "claude-3-5-sonnet-20241022",
  "messages": [
    {
      "role": "user",
      "content": [
        { "type": "text", "text": "Summarize the document content" },
        { "type": "file", "file": { "filename": "api-doc.pdf", "file_data": "https://www.bt.cn/data/api-doc.pdf" } }
      ]
    }
  ],
  "max_tokens": 1000,
  "stream": false
}

Example (PDF analysis, format 2: file_url):

{
  "model": "claude-3-5-sonnet-20241022",
  "messages": [
    {
      "role": "user",
      "content": [
        { "type": "text", "text": "Summarize the document content" },
        { "type": "file_url", "file_url": { "url": "https://www.bt.cn/data/api-doc.pdf" } }
      ]
    }
  ],
  "max_tokens": 1000,
  "stream": false
}

---

curl example

curl -X POST "https://api.gpt.ge/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxx" \
  -d '{"model":"claude-3-5-sonnet-20241022","messages":[{"role":"user","content":[{"type":"text","text":"Summarize the document content"},{"type":"file_url","file_url":{"url":"https://www.bt.cn/data/api-doc.pdf"}}]}],"max_tokens":1000}'

JavaScript (fetch) example

fetch('https://api.gpt.ge/v1/chat/completions', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json', 'Authorization': 'Bearer sk-xxxx' },
  body: JSON.stringify({ model: 'claude-3-5-sonnet-20241022', messages: [{ role: 'user', content: [{ type: 'text', text: 'Summarize the document content' }, { type: 'file_url', file_url: { url: 'https://www.bt.cn/data/api-doc.pdf' } }] }], max_tokens: 1000 })
}).then(r => r.json()).then(console.log)

Python example (requests)

import requests

payload = {
  'model': 'claude-3-5-sonnet-20241022',
  'messages': [{'role':'user','content':[{'type':'text','text':'Summarize the document content'},{'type':'file_url','file_url':{'url':'https://www.bt.cn/data/api-doc.pdf'}}]}],
  'max_tokens': 1000
}

resp = requests.post('https://api.gpt.ge/v1/chat/completions', headers={'Content-Type':'application/json','Authorization':'Bearer sk-xxxx'}, json=payload)
print(resp.json())

---

Response example (200)

{
  "id": "chatcmpl-msg_vrtx_01PjUA6WvGYYorf5rFU5QoKC",
  "model": "claude-3-5-sonnet-20240620",
  "object": "chat.completion",
  "created": 1724998217,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! I am an AI assistant named Claude. I was developed by Anthropic to converse with humans, answer questions, and provide help. Nice to meet you! How can I assist you today?"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": { "prompt_tokens": 14, "completion_tokens": 86, "total_tokens": 100 }
}

---

Server example: https://api.gpt.ge