// the find
Kav-K/GPTDiscord
A robust, all-in-one GPT interface for Discord. ChatGPT-style conversations, image generation, AI-moderation, custom indexes/knowledgebase, youtube summarizer, and more!
GPTDiscord is a self-hosted Discord bot that wraps OpenAI's API into slash commands: ChatGPT-style conversations, DALL-E image generation, a code interpreter sandbox via E2B, document/RAG indexing, AI moderation, and translation. It's for server admins who want a full-featured GPT bot in their Discord without building one from scratch.
Genuinely broad feature surface — multi-modal chat, code execution, RAG over your own docs, moderation — all exposed as real Discord slash commands rather than a single chat window bolted on. Built async from the start with a message queue service and automatic pagination for long responses, which matters for a bot meant to serve many users on one process. Ships as both a PyPI package and a Docker image with CI for build/publish, plus separate guides for Docker, a DigitalOcean droplet, and Azure App Service, so self-hosting isn't guesswork.
It's one process trying to be chat bot, code sandbox, image generator, moderator, and search engine at once — the cogs/ split helps but a lot of shared state likely still runs through openai_model.py. The README itself admits it's mid-migration from Pinecone to Qdrant, so anyone deploying now has to figure out which vector backend path is actually current. Commands still reference retired models like Davinci-003 and curie-001 next to GPT-4-Turbo and DALL-E 3, a sign the command surface hasn't kept pace with OpenAI's deprecations. Test coverage is essentially one file (tests/test_requests.py) for a codebase integrating half a dozen external APIs (OpenAI, Google, Wolfram, DeepL, E2B) — that's thin given how much can silently break when any one of those changes.