gpt4tg

This project is a Telegram bot for interactive conversations with advanced language models (OpenAI, Gemini, etc). The bot can be added to Telegram groups and supports chat-based communication, group activation, and custom usernames for users. It supports LLM session management and group-specific permissions.

Initial Setup

1. Clone the repository: Clone this repository using git clone.
2. Create Virtual Env: Create a Python Virtual Environment venv to download the required dependencies and libraries.
3. Download Dependencies: Download the required dependencies into the Virtual Environment venv using uv.

git clone https://github.com/ergolyam/gpt4tg.git
cd gpt4tg
python -m venv .venv
.venv/bin/python -m pip install uv
.venv/bin/python -m uv sync

Usage

Deploy

• Run the bot:

  TG_TOKEN="telegram_bot_token" .venv/bin/python bot

Container

• Pull the container:

  podman pull ghcr.io/ergolyam/gpt4tg:latest

• Deploy using Podman:

  mkdir -p $HOME/database/ && \
  podman run -d \
  --name gpt4tg \
  -v $HOME/database:/app/database:z \
  -e TG_TOKEN="your_telegram_bot_token" \
  ghcr.io/ergolyam/gpt4tg:latest

Environment Variables

The following environment variables control the startup of the project:

Variable	Values	Description
LOG_LEVEL	DEBUG, INFO, WARNING, ERROR	Logging verbosity
TG_ID	integer	Telegram API ID from my.telegram.org
TG_HASH	string	Telegram API hash
TG_TOKEN	string	Bot token issued by @BotFather
DB_PATH	string	Path to SQLite database file (default data.db)
MODEL_NAME	string	Name of the LLM model to use (gpt-3.5-turbo, gemini-2.0-flash)
API_KEY	comma-separated	API keys for the LLM providers
OPENAI_BASE_URL	string	Custom OpenAI API base URL (optional)
HISTORY_LIMIT	integer	Size of the message history in each session (default 10)

Basic Bot Commands

• /gpt <your_message> — Start a conversation with the bot.
• /gpt {system prompt} <your_message> — Set a custom prompt for this conversation.
• /gpt !setname <username> — Set or change your chat username.
• /gpt !setname reset — Reset your custom username.
  • The bot requires activation in each group before use. The activation password is generated and logged when the bot joins a group. Use /activate <password> to enable the bot.

Features

• AI-powered Conversations: Users can interact with advanced language models directly in Telegram groups.
• Group Activation System: Bot can be enabled/disabled in each group via activation password (for privacy and moderation).
• Username Management: Users can set custom usernames or reset them with chat commands.
• Multi-Provider Support: Works with both OpenAI and Gemini (Google) APIs, supports multiple API keys and fallback logic.
• Session Management: Each message chain maintains its own chat session context for relevant replies.