RAG API

Please refer to the Quick Start Guide to obtain your API key.

Replace $YOUR_API_KEY with the actual API key you generated in the previous step.
Our platform enforces data isolation based on API keys. Even under the same user account, data accessed through different API keys (tokens) is completely independent and not shared between them.

---

Document Management Interface

Method	Route	Function Description	API Interface
POST	/documents/upload	Processes document upload requests, including file saving, text chunking, and data persistence	https://api-platform.ope.ai/documents/upload
GET	/documents/documents	Gets a list of all documents, including document ID, filename, upload time, number of chunks, etc., with query results sorted by upload time in descending order	https://api-platform.ope.ai/documents/documents
DELETE	/documents/documents	Deletes specified documents and related resources, including: physical file deletion, Milvus vector data deletion, database record deletion, BM25 index deletion	https://api-platform.ope.ai/documents/documents
GET	/documents/document_chunks	Gets a list of all chunks for a specified document	https://api-platform.ope.ai/documents/document_chunks
GET	/documents/download_document	Downloads a specified document file	https://api-platform.ope.ai/documents/download_document

---

Database Management Interface

Method	Route	Function Description	API Interface
POST	/database/collections	Creates a user space collection. This interface is used to create a new Milvus Collection for the specified user to store vector data; returns error if creation fails or cannot connect to Milvus	https://api-platform.ope.ai/database/collections
DELETE	/database/collections	Deletes a user space collection. This interface is used to delete the specified user's Milvus Collection and all vector data stored within it	https://api-platform.ope.ai/database/collections

---

Search Interface

Method	Route	Function Description	API Interface
POST	/search/chat_message	RAG-based conversational search, processes RAG search requests Receives user questions and relevant parameters, uses RAG (Retrieval-Augmented Generation) method to generate answers. The entire process includes: - Document retrieval: Retrieving relevant documents from the knowledge base - Context integration: Combining retrieved documents with the question - Answer generation: Using a large language model to generate responses - Streaming response: Real-time return of generated content	https://api-platform.ope.ai/search/chat_message

---

Session Management Interface

Method	Route	Function Description	API Interface
POST	/session_manage/delete	Deletes a specified user session, i.e., this interface is used to manually specify a user's specific session and all related chat history	https://api-platform.ope.ai/session_manage/delete

---

API Request Examples

/documents/upload

POST Upload Document

curlPython

curl --location --request POST 'https://api-platform.ope.ai/documents/upload' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive' \
  --header 'Content-Type: multipart/form-data; boundary=------------------------897026663741264421313454' \
  --form 'files=@"/Users/ten/Documents/test.txt"' \   # Supported file types: .txt, .pdf, .csv, .md, .docx
  --form 'model="opeai/bge-m3"'                       # Embedding model name
  --form 'chunk_size=500'                             # Document chunking token length
  --form 'chunk_overlap=50'                           # Document chunking overlap length
  --form 'separators=!'                               # Document chunking separator, defaults to \n\n if not specified

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Accept': '*/*',
}

files = {
    'files': open('/Users/ten/Documents/test.txt', 'rb'),       # Supported file types: .txt, .pdf, .csv, .md, .docx
    'model': (None, 'opeai/bge-m3'),                            # Embedding model name
    'chunk_size': (None, '500'),                                # Document chunking token length
    'chunk_overlap': (None, '50'),                              # Document chunking overlap length
    'separators': (None, '!'),                                  # Document chunking separator, defaults to \n\n if not specified
}

response = requests.post('https://api-platform.ope.ai/documents/upload', headers=headers, files=files)
for line in response.iter_lines():
    if line:
        decoded = line.decode('utf-8')
        print(decoded)

/documents/document_chunks

GET Retrieve Chunk List of a Specific Document

curlPython

curl --location --request GET 'https://api-platform.ope.ai/documents/document_chunks?doc_id=14ba616e-2214-412e-8d3e-9164970de543' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Accept: */*' \
  --header 'Host': 'api-platform.ope.ai' \
  --header 'Connection: keep-alive'
  # doc_id is returned from /documents/documents

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

params = {
    'doc_id': '14ba616e-2214-412e-8d3e-9164970de543',  # doc_id is returned from /documents/documents
}

response = requests.get('https://api-platform.ope.ai/documents/document_chunks', params=params, headers=headers)
print(response.json())

/documents/download_document

GET Download Document

curlPython

curl --location --request GET 'https://api-platform.ope.ai/documents/download_document?doc_file_path=storage/docs/158/14ba616e-2214-412e-8d3e-9164970de543_test.txt&filename=test.txt' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive'
  # filename is the name used to save the file locally
  # doc_file_path is returned from /documents/documents

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

params = {
    'doc_file_path': 'storage/docs/158/14ba616e-2214-412e-8d3e-9164970de543_test.txt',  # doc_file_path is returned from /documents/documents
    'filename': 'test.txt'                                                              # filename is the name used to save the file locally
}

response = requests.get(
    'https://api-platform.ope.ai/documents/download_document',
    headers=headers,
    params=params
)

with open(params['filename'], 'wb') as f:
    f.write(response.content)

/database/collections

POST Create Knowledge Base for the User

curlPython

curl --location --request POST 'https://api-platform.ope.ai/database/collections' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive' \
  --data-raw ''
  # Use this first to create the user's knowledge base

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Content-Type': 'application/json',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

response = requests.post('https://api-platform.ope.ai/database/collections', headers=headers)
print(response.json())
# Use this first to create the user's knowledge base

DELETE Delete User Knowledge Base

curlPython

curl --location --request DELETE 'https://api-platform.ope.ai/database/collections' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive' \
  --data-raw ''
  # Deletes the user's knowledge base

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Content-Type': 'application/json',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

response = requests.delete('https://api-platform.ope.ai/database/collections', headers=headers)
print(response.json())
# Deletes the user's knowledge base

/search/chat_message

POST Process RAG Search Request

curlPython

curl --location --request POST 'https://api-platform.ope.ai/search/chat_message' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive' \
  --data-raw '{
    "embedding_model_name": "opeai/bge-m3",          # Embedding model
    "llm_model_name": "opeai/deepseek-v3",           # Large language model
    "question": "什么是RAG?",                         # User question
    "session_id": "session_1",                       # Session ID; leave empty to create a new session. The API will return a new session ID. Use this ID in subsequent requests to maintain history.
    "top_k": 5,                                      # Number of most relevant documents to return from the knowledge base
    "stream": false                                  # Defaults to true
  }'

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Content-Type': 'application/json',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

payload = {
    'embedding_model_name': 'opeai/bge-m3',          # Embedding model
    'llm_model_name': 'opeai/deepseek-v3',           # Large language model
    'question': '什么是RAG?',                         # User question
    'session_id': 'session_1',                       # Session ID; leave empty to create a new session. The API will return a new session ID. Use this ID in subsequent requests to maintain history.
    'top_k': 5,                                      # Number of most relevant documents to return from the knowledge base
    'stream': False,                                 # Defaults to true
}

response = requests.post('https://api-platform.ope.ai/search/chat_message', headers=headers, json=payload)
print(response.json())

/session_manage/delete

DELETE Delete Session by Tenant

curlPython

curl --location --request POST 'https://api-platform.ope.ai/session_manage/delete' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive' \
  --data-raw '{
    "session_id": "session_1"                      # Session ID returned by chat_message
  }'

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Content-Type': 'application/json',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

payload = {
    'session_id': 'session_1',                   # Session ID returned by chat_message
}

response = requests.post('https://api-platform.ope.ai/session_manage/delete', headers=headers, json=payload)
print(response.json())

/documents/documents

DELETE Delete Specific Document

curlPython

curl --location --request DELETE 'https://api-platform.ope.ai/documents/documents' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive' \
  --data-raw '{
    "doc_id": "123"                            # Can be obtained from documents/documents
  }'
  # Delete user document

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Content-Type': 'application/json',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

payload = {
    'doc_id': '123',                          # Can be obtained from documents/documents
}

response = requests.delete('https://api-platform.ope.ai/documents/documents', headers=headers, json=payload)
print(response.json())
# Delete user document

GET List All Documents

curlPython

curl --location --request GET 'https://api-platform.ope.ai/documents/documents' \
  --header 'Authorization: Bearer $YOUR_API_KEY' \
  --header 'Accept-Language: zh-cn' \
  --header 'Accept: */*' \
  --header 'Host: api-platform.ope.ai' \
  --header 'Connection: keep-alive'
  # Get document list

import requests

headers = {
    'Authorization': 'Bearer $YOUR_API_KEY',
    'Accept-Language': 'zh-cn',
    'Accept': '*/*',
    'Host': 'api-platform.ope.ai',
    'Connection': 'keep-alive',
}

response = requests.get('https://api-platform.ope.ai/documents/documents', headers=headers)
print(response.json())
# Get document list

API Request Errors

Common errors that may occur during API calls:

404 Not Found

The API path is incorrect or does not exist.
Solution: Verify the endpoint URL. If correct but the issue persists, contact the platform administrator.

422 Parameter Error

The client request does not meet API requirements (e.g., missing fields, format issues, typo in field names).
Solution: Check the request body structure and ensure all required fields are present and correctly formatted.

500 Internal Server Error

The server encountered an unexpected condition (e.g., system crash, DB or vector engine connection failure).
Solution: Try again later. If the problem occurs repeatedly, contact the platform administrator.