English (🇺🇸) | Русский (🇷🇺)
This project allows you to launch your Telegram bot in a few minutes to communicate with free and paid AI models via OpenRouter, or local LLMs, for example, via LM Studio.
Note
This repository is a fork of the openrouter-gpt-telegram-bot project, which adds new features (such as switch current model and Markdown
formatting in bot responses) and optimizes the container startup process.
-
Register with OpenRouter and get an API key.
-
Create your Telegram bot using @BotFather and get its API token.
-
Get your telegram id using @getmyid_bot.
Tip
When you launch the bot, you will be able to see the IDs of other users in the log, to whom you can also grant access to the bot in the future.
To run locally on Windows or Linux system, download the pre-built binary (without dependencies) from the releases page.
- Create a working directory:
mkdir openrouter-bot
cd openrouter-bot
- Create
.env
file and fill in the basic parameters:
# OpenRouter api key
API_KEY=
# Free modeles: https://openrouter.ai/models?max_price=0
MODEL=deepseek/deepseek-r1:free
# Telegram api key
TELEGRAM_BOT_TOKEN=
# Your Telegram id
ADMIN_IDS=
# List of users to access the bot, separated by commas
ALLOWED_USER_IDS=
# Disable guest access (enabled by default)
GUEST_BUDGET=0
# Language used for bot responses (supported: EN/RU)
LANG=EN
The list of all available parameters is listed in the .env.example file
- Run a container using the image from Docker Hub:
docker run -d --name OpenRouter-Bot \
-v ./.env:/openrouter-bot/.env \
--restart unless-stopped \
lifailon/openrouter-bot:latest
The image is build for amd64
and arm64
(Raspberry Pi) platforms using docker buildx.
git clone https://github.com/Lifailon/openrouter-bot
cd openrouter-bot
docker-compose up -d --build