PriceAI: AI Subscription Price Comparison Radar

TL;DR

TL;DR: PriceAI is an AGPL-3.0 price comparison radar built with Next.js and Supabase that aggregates 100+ Chinese reseller channels for AI subscriptions such as ChatGPT, Claude, Gemini, and Grok, and presents the lowest in-stock price per standard SKU with a clear jump-to-source link.

Source and Accuracy Notes

• Project page: priceai.cc
• Source repository: github.com/physics-dimension/PriceAI
• License: AGPL-3.0 (verified via GitHub API license.spdx_id)
• Source last checked: 2026-06-16

What Is PriceAI?

PriceAI is the answer to a specific Chinese-market buying problem. The same Gemini Pro, ChatGPT Plus, or Claude Pro can be bought at the official site, at a regional price, with a student or device entitlement, or through reseller channels on Xianyu, card-shop sites, Telegram groups, or personal collection links. The price gap between the official price and a low-cost reseller quote can be large, but the channels are scattered, the listings are inconsistently titled, and the inventory is unreliable.

PriceAI aggregates the reseller channels, normalizes the listing titles into standard SKUs such as ChatGPT Plus, Claude Pro, or Gemini Pro, and surfaces the lowest in-stock price per SKU with the original channel, the original listing title, the last update time, and a jump-to-source link. The site does not sell anything, take payments, or guarantee any channel’s after-sales support. It is positioned as a price radar that helps a buyer open fewer pages and avoid basic information asymmetry.

Repo-Specific Setup Workflow

The project is a Next.js application backed by Supabase. The README’s online section routes users to priceai.cc for the hosted version and walks through the comparison pages for ChatGPT Plus/Pro/Team, Claude Pro, Gemini Pro, Super Grok, and the model API channels. For a local install, follow the standard Next.js + Supabase flow:

git clone https://github.com/physics-dimension/PriceAI.git
cd PriceAI
npm install
cp .env.example .env.local
npm run dev

The .env.local file needs the Supabase URL and keys, plus any optional Google Analytics 4 credentials. Without Supabase, the price listings, channel submissions, and admin flows will not work.

Deeper Analysis

The standard SKU model

The most useful design choice in PriceAI is the standard SKU model. Reseller listings are titled inconsistently. A “Plus 土区直充”, “GPT PLUS 月卡”, and “Plus 成品号” might all describe the same kind of product. PriceAI normalizes the title to a ChatGPT Plus standard SKU, keeps the original channel, original title, and original price, and shows the lowest in-stock price per SKU. The normalization is what makes the comparison usable.

In-stock lowest price

The product rule is “in-stock lowest price first.” Out-of-stock listings are not promoted to the top of the comparison view. The user can still see the full raw quote view, but the headline price for a SKU only counts quotes that are currently marked as in stock. That is the difference between a price radar and a misleading low-price headline.

Channel submission loop

The README documents a channel submission flow. Users can submit a new reseller channel through the front end. The admin flow then runs a test scrape, adds the scraper to the collector backlog, and starts aggregating quotes. The loop is what keeps the channel list growing without a manual content team.

Aggregation strategies

The collector supports public API scraping, Shop API scraping, HTML parsing, and a browser-based fallback. The four strategies let the system ingest a wide range of reseller site shapes without a per-site rewrite. The README’s tech notes mention a Supabase backend and a Next.js + TypeScript frontend, with a Node.js-based collector.

What the platform does not do

The README is explicit about the limits. PriceAI does not sell products, take payments, or guarantee any reseller’s after-sales support. The final purchase risk stays with the user. The platform is the comparison layer, not the merchant layer, and the user’s own judgment about channel reliability is the deciding factor.

Use cases

The README’s “适合谁” (who it’s for) section lists four reader profiles: people who buy AI subscriptions, accounts, or quotas regularly; people who want the best price without doing the channel research themselves; users with limited overseas payment options; and power users who already have a personal collection of card-shop or group links and want a unified panel. The use cases all share a common trait: the reader wants the price and the source link in one place, not a 20-tab comparison session.

Practical Evaluation Checklist

• [ ] Do you buy ChatGPT, Claude, Gemini, or Grok subscriptions through Chinese reseller channels?
• [ ] Do you want the lowest in-stock price per standard SKU rather than a single raw quote?
• [ ] Will you use the channel submission flow to expand the aggregator?
• [ ] Do you have a Supabase project and the URL plus keys to wire up the backend?
• [ ] Are you comfortable with AGPL-3.0 if you want to self-host a private fork?
• [ ] Do you need Google Analytics 4 for the access analysis view?
• [ ] Will you keep the comparison view’s “in-stock only” rule intact in any private fork?

Security Notes

PriceAI is a Next.js + Supabase app, so the standard rules apply. The Supabase URL and service role key are the most sensitive secrets. Keep them out of version control and rotate the service role key on a schedule. The admin flow exposes channel management, batch scraping, quote hiding, and rebuild actions, so the admin path needs the same access controls you would use for any other internal tool.

The platform does not sell products, take payments, or guarantee reseller support. The user is the final risk owner for any purchase decision. If you use the channel submission flow, treat the submitted channel as untrusted until the admin flow has verified the scrape and the inventory status. A misconfigured scraper can produce phantom in-stock prices that mislead users.

FAQ

Q: Does PriceAI sell AI subscriptions? A: No. PriceAI is a price comparison radar. It does not sell products, take payments, or guarantee any reseller’s after-sales support.

Q: How does the lowest price get picked? A: The headline price for a standard SKU is the lowest quote currently marked as in stock. Out-of-stock quotes are not promoted to the top of the comparison view.

Q: What platforms and SKUs are supported? A: ChatGPT, Claude, Gemini, Grok, API/CDK, mailboxes, and “other” categories. The “other” bucket covers verification-receiving services, virtual cards, branded tool accounts, and unlisted tool accounts.

Q: Can users submit new channels? A: Yes. The channel submission flow is documented in the README. Submissions go through an admin flow with a test scrape and a collector backlog, which is what keeps the channel list growing.

Q: What license is PriceAI released under? A: AGPL-3.0, verified via the GitHub API. A private fork for personal use is allowed under the AGPL; a hosted fork that you expose to other users is subject to the AGPL’s source-sharing terms.

Conclusion

PriceAI is a focused comparison radar for a market that does not have a clean aggregator. The standard SKU model and the in-stock lowest price rule are the two design choices that make the comparison usable. The AGPL-3.0 license is a real constraint for a hosted fork, but the open-source model is what makes the channel submission loop viable. If you buy AI subscriptions through Chinese reseller channels, PriceAI is a useful page to keep open. If you are outside that market, the project’s comparison patterns are still worth studying before you build your own.

Related reading: GitHub Trending tools, Developer tools, SEO Score API, Klaus, API Radar.