Run the EveryAPI gateway inside NextChat
NextChat is a self-hostable chat interface. Point its OpenAI-compatible configuration at EveryAPI, then add the model IDs you want NextChat to show.
One base URL line points NextChat at EveryAPI
Default global Base URL / China-first Base URL: https://api.everyapi.ai/v1 / https://api-cn.everyapi.ai/v1
# Set these environment variables for a Vercel deployment OPENAI_API_KEY sk-everyapi-...BASE_URL https://api.everyapi.ai # CUSTOM_MODELS additions require a + prefixCUSTOM_MODELS +claude-sonnet-5,+gpt-5.5,+gemini-3.1-pro-low,+MiniMax-M3From zero to your first call
- Deploy NextChat in one click
We recommend Vercel's one-click deploy (the button is in the GitHub repo README); you can also run docker run yidadaa/chatgpt-next-web.
github.com/ChatGPTNextWeb/NextChat
- Set env vars at deploy time or fill them in at runtime
Setting BASE_URL=https://api.everyapi.ai + OPENAI_API_KEY=sk-everyapi-... at deploy time is the site-wide default; the runtime Settings → Custom Endpoint is configured per browser session.
- Customize the model list
Add model IDs with the required + prefix, for example +claude-sonnet-5,+gpt-5.5. Only add IDs that appear in your EveryAPI account catalog.
- Start chatting
Return to the main view, select one of the configured models, and send a test message. That model entry uses the EveryAPI endpoint you configured.
NextChat × EveryAPI
Why doesn't the BASE_URL field include /v1?
NextChat's internal code appends the path segments itself (/v1/chat/completions, etc.), so the BASE_URL / endpoint field only needs the root URL https://api.everyapi.ai. This is the opposite of Cursor's "Override OpenAI Base URL", which does need /v1 — it depends on how the client builds the path.