Forma is an open-source ChatGPT bot written in Go that brings OpenAI’s models into WeChat, with Telegram support as well. Published on GitHub under the MIT license with about 846 stars and last updated in 2024, it builds on the earlier wechatgpt codebase and extends it with multimodal conversation. Deploy it with Docker, scan a QR code to log in, and your WeChat account can answer friends and groups with GPT — including looking at images they send.
Forma is configured through environment variables or a small config file: set your OpenAI API key, choose your models, and decide how the bot triggers. In WeChat it can reply to any message or only when a keyword appears, and it adds two multimodal tricks on top of plain chat — GPT-4 Vision recognition, so the bot can see and respond to images, and DALL·E image generation triggered by a keyword in chat. A Telegram mode is included with keyword triggers and a user whitelist, although the maintainer notes the Telegram side is less actively maintained. Docker Compose deployment keeps the whole thing to a few commands.
Forma’s lineage is part of its appeal: it descends from one of the earliest WeChat ChatGPT bridges and keeps that codebase’s simplicity while adding the multimodal features users asked for. Configuration is honest and file-based, and the maintainers document the common pitfalls, from API-key errors to Go version requirements, in a built-in Q&A. For a personal bot that friends can chat with and send pictures to, it remains a pragmatic, hackable choice.
Forma is completely free and open source under the MIT license. You only pay for your own OpenAI API usage.
Developers in the WeChat ecosystem who want a self-hosted, multimodal ChatGPT bot for personal or group use, with full control over triggers, models, and deployment.