chatgpt_telegram_bot is an MIT-licensed Python bot that places OpenAI's GPT models and Anthropic's Claude models inside Telegram using your own API keys, aimed at individuals and teams who want fast, uncapped model access without living in a browser tab.

What it is

chatgpt_telegram_bot is a Python application that connects a Telegram bot to OpenAI's native API and, optionally, to OpenRouter for routing Anthropic's Claude models and GPT-5.5. It is config-driven: models are declared in config/models.yml and chat modes in config/chat_modes.yml, so entries can be added or removed with no code changes. Shipped options include GPT-4o mini as the default, GPT-4o, GPT-5.5, and Claude Opus 4.8, Sonnet, and Haiku, alongside OpenAI's gpt-image-1 for image generation and Whisper for voice transcription. It belongs to the Telegram bot ecosystem and sits on top of the OpenAI and OpenRouter API surfaces.

The concrete problem it solves is the browser-based chat interface. The README describes web chatbots as laggy, rate-limited, and locked behind a browser tab. This bot replaces that client with a Telegram-native one where the operator supplies their own API keys, so requests are not capped by a daily limit and spend goes directly to the API provider rather than through an intermediary. Replies usually arrive in three to five seconds and stream into the chat word-by-word.

Key capabilities

• Message streaming that delivers answers into Telegram word-by-word.
• Fifteen chat modes, including General Assistant, Code Assistant, Artist, Psychologist, Elon Musk, SQL Assistant, and Movie Expert, with custom modes added in config/chat_modes.yml.
• Vision support, so a vision-capable model such as GPT-5.5 or Claude can read a sent image.
• Image generation through OpenAI's gpt-image-1, enabled in Artist mode.
• Voice messages transcribed by Whisper.
• A config-driven model list in config/models.yml, editable without code changes, including any other OpenRouter-routed model declared with provider: openrouter.
• Access control restricting the bot to a list of allowed Telegram users, plus balance tracking via /balance and code highlighting for formatted code blocks.

Who uses it and how

• Individuals who want a private ChatGPT or Claude client inside Telegram and are willing to hold their own OpenAI and OpenRouter API keys.
• Group chats, supported through the /help_group_chat setup instructions.
• Operators who need to limit access, using the allowed-user list to keep the bot closed to everyone else.
• Cost-conscious users who track spend with /balance and pick models by price and speed, such as GPT-4o mini as the default for cheap, fast replies.
• Self-hosters who want frontier models such as GPT-5.5 or Claude Opus 4.8 without a provider-imposed daily cap.

Getting started

Rename config/config.example.yml to config/config.yml and config/config.example.env to config/config.env, then set telegram_token from @BotFather and openai_api_key, adding openrouter_api_key for Claude and GPT-5.5, and run docker-compose --env-file config/config.env up --build. A hosted instance exists at @jadvebot and jadve.com for those who would rather not deploy their own.

How it compares

No list of paid products this project replaces is provided in the facts, and no directly comparable tool is named there either. As presented in this registry, it stands alone.

When to use it — and when not to

A self-hoster must supply a Telegram bot token from @BotFather, an OpenAI API key, and optionally an OpenRouter API key, then run the service through Docker Compose while paying the API providers directly. It is a poor fit for anyone who wants a fully managed service with no keys to hold and no API spend to monitor, although the hosted @jadvebot serves that audience. The licence is unambiguous at MIT, but the README excerpt available here is partial and does not document operational details such as persistence or scaling.

mv config/config.example.yml config/config.yml
mv config/config.example.env config/config.env
# then edit config/config.yml — set telegram_token, openai_api_key (and openrouter_api_key for Claude/GPT-5.5)

docker-compose --env-file config/config.env up --build

bot/
  bot.py           # Telegram handlers, streaming, commands
  openai_utils.py  # model dispatch (OpenAI + OpenRouter), token counting, vision
  config.py        # loads config.yml / models.yml / chat_modes.yml
  database.py      # MongoDB storage for users & dialogs
config/
  config.yml       # your tokens & settings
  models.yml       # model catalog, pricing, capabilities
  chat_modes.yml   # chat-mode prompts