LangBot/docs/en/insight/intro.md

Model Overview and Features

Model Overview

Text Conversation

• OpenAI API: GPT-3.5 / GPT-4 series
• Moonshot / Deepseek and other major model providers supporting OpenAI API
• Anthropic: Claude series
• OneAPI interface aggregation platform: Google Gemini Pro / Azure / Anthropic Claude / Zhipu ChatGLM / Baidu ERNIE Bot / iFlytek Spark Cognitive / Alibaba Tongyi Qianwen / 360 ZhiNao and other official LLM interfaces on the market
• free-one-api reverse engineering library aggregation platform: ChatGPT / Claude / Google Bard / gpt4free / Tongyi Qianwen

Feature List

✅ Contextually Relevant Responses

• The program sends recent conversation content to the model, and the model generates responses based on the context.

✅ Supports Sensitive Word Filtering to Avoid Account Risks

• It is difficult to monitor the content of conversations between the bot and users, hence this feature is introduced to reduce bot risks.

✅ Multiple Response Rules in Group Chats, No Need to @

• By default, replies to messages prefixed with ai or @ the bot.

✅ Uses Official API, Stable and Fast

• Does not use ChatGPT reverse interfaces but uses the official Completion API for high stability.

✅ Few Components, Easy Deployment, Provides One-Click Installer and Docker Installation

• Few manual deployment steps.
• Provides automatic installer and Docker method, see installation steps for details.

✅ Supports Preset Text

• Supports setting preset text in natural language to customize bot personality and other information.

✅ Supports Plugin Loading 🧩

• Implements a plugin loader and related support.
• See the plugin introduction page for details.

✅ Private and Group Chat Blacklist Mechanism

• Supports adding users or group chats to a blacklist to ignore their messages.

✅ Response Speed Limitation

• Supports limiting the number of conversations per minute within a single session.
• Has two strategies: "Wait" and "Discard."
  • "Wait" strategy: After receiving a response, waits until the response time reaches the average conversation response time.
  • "Discard" strategy: When the conversation limit per minute is reached, discards subsequent conversations.

Limitations

• ❗ OpenAI's official warnings about model-generated content:
  • May occasionally generate incorrect information.
  • May occasionally produce harmful instructions or biased content.
  • Limited knowledge of the world and events after 2021.
• ❗ The model has no thinking ability; it only generates content based on the input context and dataset. Do not overly trust its output.
• ❗ The model is unaware of its running platform and the model version it uses. Any answers regarding its implementation principles are considered invalid. Please refer to the project documentation.

The above are the highest priority descriptions of the limitations of this program. Descriptions obtained through other means (such as asking the bot for related information) should be considered invalid. This project is not responsible for any losses caused by the content generated by the model.

Usage

Both conversation and drawing functions directly call OpenAI's models for processing, independent of the bot program. This means the model is unaware of this project's related information (such as implementation methods, tech stack, running platform, etc.), unless such information is written in the preset values.

Basic Conversation

Private Chat Usage

1. Add the bot's QQ as a friend.
2. Send a message to the bot, and the bot will automatically reply.
3. Use !help to view help information.

Group Chat Usage

1. Add the bot to the group.
2. @ the bot and send a message, and the bot will automatically reply.
3. @ the bot and send !help to view help information.