The English documentation is placed below the Chinese version.

sider_ai_api是一个调用sider.ai的Python API库，提供了在不能访问ChatGPT等国外AI平台的情况下，调用ChatGPT、Gemini、Claude、llama，乃至o1等国外大模型的完整解决方案。

安装

打开终端，并输入pip install sider-ai-api即可。

用法

Session 类

Session 类用于与 Sider API 交互，提供聊天和 OCR 功能，同时管理 API 的身份验证、请求头和会话状态。

构造函数:

def __init__(self,token=None,context_id="",cookie=None):

• 参数:
  • token (str): 可选，身份验证令牌，如"eyJhbGciOiJIUzI..."，注意不需要加Bearer 这个开头。如果为None，则自动读取DEFAULT_TOKEN_FILE (默认是_token.json) 中的token和cookie。
  • context_id (str): 可选，对话上下文的 ID，如"C0M.......EB"，空字符串表示开始新对话。默认为空字符串。
  • cookie (str): 可选，身份验证所需的 cookie 字符串，如"key1=value1;key2=value2"。如果不提供，默认使用COOKIE_TEMPLATE常量中的模板。

主要方法:

chat(prompt, model="gpt-4o-mini")

用提示词与 AI 对话，返回一个字符串生成器，包含AI的响应结果。

• 参数:

  • prompt (str): 用户提示词。
  • model (str): 使用的 AI 模型，默认为 "gpt-4o-mini"。

• 返回: 一个字符串生成器，逐步返回 AI 的响应，可以使用"".join()拼接。

ocr(filename, model="gemini-2.0-flash")

对图像文件执行 OCR 识别，返回结果的字符串生成器。

• 参数:

  • filename (str): 图像文件的路径。
  • model (str): 使用的 OCR 模型，默认为 "gemini-2.0-flash"。完整的模型列表可以在MODELS和ADVANCED_MODELS常量中找到。

• 返回: 一个字符串生成器，逐步返回 AI 的响应。

属性:

• context_id: 对话上下文id，如果为""则表示开始新对话。对话上下文会由服务器端保存。
• total: API的总调用次数，为整数或者None。
• remain: API的剩余调用次数，为整数或者None。

示例代码

from sider_ai_api import Session

# 首先设置token和cookie变量
session = Session(token=token, cookie=cookie)
print("".join(session.chat("Hello, ChatGPT!", "gpt-4o-mini")))
print("".join(session.chat("Hello, Claude!", "claude-3.5-haiku")))
print("".join(session.chat("Hello, Gemini!", "gemini-2.0-flash")))
print("".join(session.chat("Hello, DeepSeek!", "deepseek-chat")))  # DeepSeek-v3
print("".join(session.chat("Hello, o1!", "o1")))
print("".join(session.ocr("path/to/image.jpg"))) # OCR
print(f"还剩{session.remain}/{session.total}次API调用。")

关于token和cookie

Session类是必须提供token的。token可以在浏览器的设置或者开发者工具中查看。对于Edge，cookie和token可以在edge://settings/cookies/detail?site=sider.ai中查看。
如果只有token而没有额外的cookie信息，如CloudFront-Signature等，也一样能调用聊天功能，但是在OCR识别的上传图片这一步，会提示需要Cloudflare验证的错误消息，因此建议包含完整的cookie。

---

sider_ai_api is a Python API library for interacting with sider.ai. It provides a complete solution for accessing large language models such as ChatGPT, Gemini, Claude, Llama, and even o1, in scenarios where direct access to foreign AI platforms like ChatGPT is unavailable.

Installation

Just open the terminal and enter pip install sider-ai-api.

Usage

Session Class

The Session class is used to interact with the Sider API. It provides chat and OCR functionalities while managing API authentication, request headers, and session states.

Constructor:

def __init__(self, token=None, context_id="", cookie=None):

• Parameters:
  • token (str): Optional. The authentication token, e.g., "eyJhbGciOiJIUzI...". Note that you don't need to prepend Bearer to the token. If set to None, it will automatically read the token and cookie from DEFAULT_TOKEN_FILE (default is _token.json).
  • context_id (str): Optional. The ID of the conversation context, such as "C0M.......EB". An empty string indicates starting a new conversation. Defaults to an empty string.
  • cookie (str): Optional. The cookie string required for authentication, e.g., "key1=value1;key2=value2". If not provided, the default COOKIE_TEMPLATE constant will be used.

Main methods:

chat(prompt, model="gpt-4o-mini")

Interact with the AI using a prompt. Returns a string generator containing the AI's response.

• Parameters:

  • prompt (str): The user prompt.
  • model (str): The AI model to use. Defaults to "gpt-4o-mini". A complete list of models can be found in the MODELS and ADVANCED_MODELS constants.

• Returns: A string generator that progressively yields the AI's response. You can use "".join() to concatenate the output.

ocr(filename, model="gemini-2.0-flash")

Perform OCR (Optical Character Recognition) on an image file. Returns a string generator containing the recognition result.

• Parameters:

  • filename (str): The path to the image file.
  • model (str): The OCR model to use. Defaults to "gemini-2.0-flash".

• Returns: A string generator that progressively yields the OCR result.

Attributes:

• context_id: The conversation context ID. If it is "", it indicates starting a new conversation. The conversation context is stored on the server side.
• total: The total number of API calls available, as an integer or None.
• remain: The remaining number of API calls, as an integer or None.

Example Code

from sider_ai_api import Session

# Firstly, set the token and cookie variables
session = Session(token=token, cookie=cookie)
print("".join(session.chat("Hello, ChatGPT!", "gpt-4o-mini")))
print("".join(session.chat("Hello, Claude!", "claude-3.5-haiku")))
print("".join(session.chat("Hello, Gemini!", "gemini-2.0-flash")))
print("".join(session.chat("Hello, DeepSeek!", "deepseek-chat")))  # DeepSeek-v3
print("".join(session.chat("Hello, o1!", "o1")))
print("".join(session.ocr("path/to/image.jpg")))  # OCR
print(f"Remaining {session.remain}/{session.total} API calls.")

About Token and Cookie

The Session class requires a token that can be found in your browser settings or developer tools. For Edge, you can view the cookie and token at edge://settings/cookies/detail?site=sider.ai.
If you only have the token and not additional cookie information (e.g., CloudFront-Signature), you can still use the chat functionality. However, for OCR image uploads, you may encounter an error requiring Cloudflare verification. Therefore, it is recommended to include the full cookie.