English | 简体中文 | 繁體中文 | 日本語 | Español | Français | Русский | Deutsch

Welcome to Telegram GPT Worker! 👋 This is an efficient Telegram bot developed in TypeScript. It supports multiple languages and AI models, deployed on Cloudflare Workers, providing users with a fast and scalable service experience.

1. 🧠 Multi-model Support: Integrates multiple AI models including OpenAI, Google Gemini, Anthropic Claude, Groq, and Azure OpenAI.
2. 🔗 OpenAI Compatible Model Support: Designed specifically for AI model interface management and distribution systems such as One API and New API, supporting automatic model list retrieval.
3. 💬 Intelligent Conversation: Equipped with context memory capability, ensuring smooth and natural dialogues.
4. 🎨 Image Generation: Supports text-to-image generation using DALL·E and Cloudflare Flux technologies.
5. 🖼️ Image Analysis: Allows users to upload images for intelligent analysis, using either OpenAI or Google Gemini models.
6. 🌍 Multilingual Support: Built-in i18n functionality, supporting 8 languages to meet diverse needs.
7. 🔒 User Permission Management: Controls access through whitelist functionality, enhancing security.
8. ☁️ High-performance Deployment: Utilizes Cloudflare Workers' edge computing capabilities for rapid response.
9. 🗄️ Efficient Data Management: Uses Redis for data caching and management, ensuring efficient processing.
10. 🔧 Flux Prompt Optimization: Optional feature that optimizes image generation prompts for Flux model through an external API.

Before getting started, please ensure you have the following:

• Cloudflare account
• Telegram account and Bot Token
• Upstash Redis database (with Eviction feature enabled)
• API key for at least one AI service

1. Clone the project repository
2. Configure necessary environment variables
3. Deploy to Cloudflare Workers
4. Set up Telegram Webhook

For detailed deployment steps, please refer to the tutorial below.

• /start - Start the bot
• /language - Switch language
• /switchmodel - Switch AI model
• /new - Start a new conversation
• /history - Get conversation history summary
• /help - Get help information
• /img - Generate image (DALL-E)
• /flux - Generate image (Cloudflare Flux)

/GPT-Telegram-Worker
│
├── /src
│   ├── /api
│   │   ├── azure.ts               # Handle Azure API interactions
│   │   ├── claude.ts              # Handle Claude API interactions
│   │   ├── flux-cf.ts             # Handle Cloudflare AI drawing interface
│   │   ├── gemini.ts              # Handle Google Gemini API interactions
│   │   ├── groq.ts                # Handle Groq API interactions
│   │   ├── image_generation.ts    # Handle DALL·E drawing interface
│   │   ├── model_api_interface.ts # Common interface defining model API standard structure
│   │   ├── openai_api.ts          # Handle OpenAI API interactions
│   │   ├── openai_compatible.ts   # Handles OpenAI compatible API interactions
│   │   └── telegram.ts            # Handle Telegram bot logic
│   ├── /config                    # Configuration files
│   │   └── commands.ts            # Telegram bot commands
│   ├── /utils
│   │   └── helpers.ts             # Utility functions and tools
│   │   └── i18n.ts                # Multilingual functions
│   │   └── redis.ts               # Upstash Redis functions
│   │   └── image_analyze.ts       # Image upload functions
│   ├── index.ts                   # Entry file, handling requests and responses
│   └── env.ts                     # Configure environment variables
├── /types                         # Type definition files
│   └── telegram.d.ts              # Type definitions for Telegram API
├── wrangler.toml                  # Cloudflare Worker configuration file
├── tsconfig.json                  # TypeScript configuration file
├── package.json                   # Project dependency file
└── README.md                      # Project documentation

1. Install Wrangler CLI:

   npm install -g @cloudflare/wrangler

2. Log in to your Cloudflare account:

   wrangler login

3. Create a new Workers project:

   wrangler init telegram-bot

4. Copy the dist/index.js file into your project.

5. Edit the wrangler.toml file to configure your project:

   name = "telegram-bot"
   type = "javascript"
   account_id = "your_account_id"
   workers_dev = true

6. Deploy to Cloudflare Workers:

   wrangler publish

1. Log in to the Cloudflare Dashboard.
2. Select "Workers & Pages".
3. Click "Create application" and choose "Create Worker".
4. Name your Worker and click "Deploy".
5. Copy and paste the contents of dist/index.js into the editor, then save the file.
6. Add necessary environment variables in "Settings".

Use the Telegram Bot API to set up the Webhook. URL example:

https://api.telegram.org/bot<YOUR_BOT_TOKEN>/setWebhook?url=https://your-worker.your-subdomain.workers.dev/webhook

https://api.telegram.org/bot123456789:abcdefghijklmn/setWebhook?url=https://gpt-telegram-worker.abcdefg.workers.dev/webhook

1. Clone the project:

   git clone https://github.com/snakeying/telegram-bot.git

2. Install dependencies:

   npm install

3. Set up environment variables.

4. Compile TypeScript:

   npm run build

5. Start the bot:

   npm start

Note: Some variables need to be manually configured and have no default values.

Allows users to upload images and receive AI analysis results. Here's how to use it:

1. User sends an image to the bot.
2. In the image caption, add an analysis prompt, e.g., "Please analyze this image".
3. The bot will use the currently selected AI model (OpenAI or Google Gemini) to analyze the image.
4. The analysis result will be returned to the user as a text message.

Note: Make sure the AI model you're using supports image analysis. If the current model doesn't support it, the bot will prompt you to switch to a multimodal-supporting model.

When the PROMPT_OPTIMIZATION environment variable is set to true, the Flux image generation feature uses an external API to optimize prompts. This feature works through the following steps:

1. User provides the original prompt.
2. The system uses the external API configured with EXTERNAL_API_BASE, EXTERNAL_MODEL, and EXTERNAL_API_KEY to optimize the prompt.
3. The optimized prompt is used by the Flux model to generate the image.

This feature can help generate more precise images that better align with Flux model characteristics. To use this feature, make sure all relevant environment variables are correctly configured.

1. 🚦 Use API Quotas Responsibly: Be mindful of usage limits, especially for image generation and analysis services.
2. 🔐 Protect Sensitive Information: Safeguard your environment variables and API keys.
3. 🧠 Understand Model Characteristics: Choose the AI model that best fits your application scenario.
4. 🔄 Stay Updated: Regularly update code and features for optimal performance.
5. 🛡️ Security First: Update API keys periodically and follow the principle of least privilege.
6. 🎨 Flux Prompt Optimization: When enabling PROMPT_OPTIMIZATION, ensure correct configuration of EXTERNAL_API_BASE, EXTERNAL_MODEL, and EXTERNAL_API_KEY.
7. ⛔ Important Notice: To avoid potential conflicts, it is not recommended to add models already used by other APIs in OpenAI Compatible. For instance, if you have configured the Gemini API and selected the gemini-1.5-flash model, you should not add the same model in OpenAI Compatible.

• Bot not responding? Check Webhook settings and environment variable configurations.
• Encountering API limits? Review your API quota usage.
• Image analysis failing? Ensure you're using a multimodal-supporting model like GPT-4o/GPT-4o-mini or Gemini 1.5 Pro/flash.

This project is licensed under the MIT License.

Copyright (c) 2024 [snakeying]