OpenGPTs is an open-source project created to create flexible and highly customizable chatbots similar to OpenAI’s GPTs. Built on top of LangChain, LangServe, and LangSmith, it provides more control over the large language models, prompts, and tools you can use to build AI assistants.
GPTs are a new way for anyone to create a tailored version of ChatGPT to be more helpful in their daily life, at specific tasks, at work, or at home—and then share that creation with others.
From OpenAI
OpenGPTs brings several benefits over using OpenAI APIs directly. The main advantage is the ability to customize the language model powering your chatbot easily.
With OpenGPTs’ integration of LangChain’s catalog of over 60 large language models, you can choose different foundations optimized for certain tasks or genres. For example, models specialized in summarization versus dialogue. This flexibility allows tailoring the chatbot to your specific needs.
Additionally, OpenGPTs facilitates customizing prompts and tools through its sandbox environment and support for adding custom actions and knowledge files. The drafted sharing and publishing features also simplify deploying and distributing your finished chatbot.
For developers wanting more control over building AI assistants, OpenGPTs provides welcome customizability versus OpenAI alone.
Here is an example App that you can create and chat with OpenGPTs.
How to use it:
1. Install requirements.
cd backend pip install -r requirements.txt
2. Set environment variables for the AI provider you want to use. Default agent types:
- “GPT 3.5 Turbo”
- “GPT 4”
- “Azure OpenAI”
- “Claude 2”
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY=sk-...
export OPENAI_API_BASE=... export OPENAI_API_VERSION=... export OPENAI_API_KEY=... export OPENAI_DEPLOYMENT_NAME=...
3. Set up LangSmith for debugging, logging, and monitoring. OPTIONAL.
export LANGCHAIN_TRACING_V2="true" export LANGCHAIN_API_KEY=...
4. Start the backend server:
langchain serve --port=8100
5. Start the frontend:
cd frontend yarn yarn dev
6. Visit http://localhost:5173/ to see the magic.
