What it does

gpt-code-ui is an open source project that you install with pip and start from a terminal with the gptcode command. You type a request in the interface, an OpenAI model generates code for it, and the code runs in a Python kernel. The interface supports file upload and download, keeps the context of your earlier messages, and lets you switch between GPT-3.5 and GPT-4. The repository suggests installing a set of data and PDF libraries, such as pandas, matplotlib and pdfplumber, in the Python environment used to run it. Your OpenAI key can be placed in a .env file in the working directory. The ports and the OpenAI endpoint can be overridden through environment variables, and Azure OpenAI Services is supported with extra settings, although choosing a model in the interface is not supported there. A Docker bundle made by the community, gpt-code-ui-docker, is linked from the page.

Capabilities

Generates code from a plain request and runs it in a Python kernel
File upload and file download
Keeps context so it can refer to your earlier messages
Switch between GPT-3.5 and GPT-4
OpenAI key read from a .env file, with a configurable endpoint and Azure OpenAI settings

When to use it

You want a code-running chat interface that works with your own OpenAI API key
You need to upload files, have code written against them and download the results
You want to point the tool at a custom or Azure OpenAI endpoint