NextChat (ChatGPT Next Web): Docker and VPS setup

This Chinese-language tutorial is summarized from its complete original Chinese captions. Uploaded on April 10, 2023, it explains how to deploy ChatGPT Next Web, the earlier name of NextChat, as a self-hosted browser interface for OpenAI's API. The demonstrated default model is GPT-3.5 Turbo, and the speaker also points out a GPT-4 option. The source gives no application release number. Its setup steps and interface reflect that period.

The speaker introduces Vercel deployment before walking through a VPS installation with Docker Compose. The configuration includes an OpenAI API key, an access password and a host port, with 8090 used in the example. He recommends setting a password to prevent visitors from consuming the owner's API allowance. Nginx Proxy Manager handles the custom domain, reverse proxy and SSL certificate.

Server access is a practical concern in the demonstration. A connectivity check returns "IP is blocked", but a later chat request succeeds on the same machine. That result does not establish that other server addresses will work. Hosting this interface also leaves model requests dependent on OpenAI's service; the tutorial does not demonstrate local inference.

The interface walkthrough covers dark mode, language and font settings, plus a request for a Linux backup script. The speaker describes Markdown chat exports and context compression. Built-in prompts include a simulated Linux terminal, while a translation prompt fails to produce the expected response during the demonstration.