What it does

This is a commercial fork of ChatGPT-Next-Web that layers on Midjourney (standard and “plus” with face-swap and inpainting), Stable Diffusion, DALL-E 3, Suno music, Luma video, and FastGPT knowledge-base integration. It comes in two flavors: a simpler “no-backend” version that runs in Docker with an authorization code, and a “with-backend” version that adds user registration, subscription billing, admin dashboards, and payment gateways (WeChat Pay, Yizhifu, Hupijiao).

The interesting bit

The project is unapologetically not open source — the author sells authorization codes and source access via WeChat — yet it keeps merging upstream ChatGPT-Next-Web changes. The backend version is essentially a turnkey SaaS kit for running a Chinese AI service, down to the QR-code WeChat login and OSS/COS/MinIO storage adapters.

Key highlights

• Multi-modal support: GPT-4V, Whisper, TTS, GPTs store, and “gpt-4-all” reverse models
• Drawing panel with Midjourney-plus (face swap, regional redraw), Stable Diffusion webui, gpt-image-1
• File parsing and S3-compatible storage (Aliyun OSS, Tencent COS, MinIO, R2) in newer versions
• Backend admin: model/API key management, subscription packages, redemption codes, order tracking, session logging
• Docker-only deployment; no-backend version has a free tier (3.8.7) and a paid tier (3.9.18)

Caveats

• Source code is not publicly available; security claims are stated but unverifiable
• Authorization codes and source access require contacting the author on WeChat (no automated licensing)
• Default admin credentials are admin/123456 — the README literally warns you to change them immediately

Verdict

Worth a look if you need a pre-integrated, Chinese-market-ready AI chat SaaS without building the plumbing yourself. Skip it if you require open-source auditability or prefer to wire together your own stack from composable parts.