NextChat Custom OpenAI API Base URL Setup

NextChat is often used as a self-hosted web or desktop AI assistant. Its server environment variables include OPENAI_API_KEY and BASE_URL, which makes it suitable for OpenAI-compatible API gateways.

Who this is for

• You self-host NextChat.
• You want OpenAI requests to go through a custom API gateway.
• You need a shared entry point for yourself or a team.
• You want one key to manage multiple models.

Core environment variables

A typical setup looks like this:

OPENAI_API_KEY=sk-your-api-key
BASE_URL=https://api.cheng-zi-ai.com

In NextChat's README, BASE_URL overrides the OpenAI API request base URL, whose default is https://api.openai.com. In most self-hosted setups, this should be the root URL rather than the /v1 endpoint used by some local clients.

Redeploy after changing variables

If you update environment variables on Vercel, Zeabur, or another platform, redeploy the app. Saving variables without redeploying may leave the app using old settings.

Model names

The visible model list may come from configuration, built-in defaults, or server-side forwarding logic. Start with a known available model:

gpt-5.3-codex

If you use Claude or Gemini models through an API gateway, make sure the gateway exposes them as OpenAI-compatible model IDs.

Common errors

The app loads, but requests fail

Confirm that server-side environment variables are active. Self-hosted apps often run with old variables after a UI-only update.

401

Check that OPENAI_API_KEY is complete, not truncated by the deployment platform, and not wrapped in extra quotes.

404

Check the BASE_URL. NextChat commonly expects the root URL. If /v1 fails, try the root URL.

Practical advice

NextChat is closer to a deployed app than a local client like Cline. Keep a small change log for key, Base URL, and model-list edits so you can roll back quickly.