sstxww/yangmao-fuli-zhongzhuanzhan — explained in plain English
Analysis updated 2026-05-18
Find free or low cost AI API relay sites organized by risk level in Chinese.
Check whether a listed relay site is quietly serving a downgraded or fake model before relying on it.
Report a broken, overcharging, or suspicious relay site so it gets reviewed and possibly blacklisted.
Submit a newly discovered relay site for community review and listing.
| sstxww/yangmao-fuli-zhongzhuanzhan | blader/arbitrage | energypantry/agent-browser-runtime | |
|---|---|---|---|
| Stars | 86 | 86 | 86 |
| Language | — | — | JavaScript |
| Setup difficulty | easy | easy | hard |
| Complexity | 1/5 | 2/5 | 4/5 |
| Audience | general | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
This repository is the documentation and landing page for a Chinese language website called welfare.sstxww1.asia, which curates so called wool hunting or freebie sites: collections of free or low cost resource entry points, along with contribution and risk reporting tools. The repository itself contains no source code, no backend passwords, no server configuration, and no database backups. Only the website itself hosts and maintains that data. The website organizes listed sites into separate zones: a free zone for resources anyone can use with little barrier, a semi paid zone for sites offering a free tier alongside paid upper limits, a reverse zone for sites that can be used directly without extra setup, and a blacklist zone for sites found to have clear problems such as being unreachable, behaving strangely, or charging unexpected fees. A detection system checks each listed site's API compatibility, tries to identify which underlying AI model actually answers requests, and runs several probes: ordinary question answering, math, JSON formatting, translation, and instruction following, to judge whether a site is quietly serving a different or downgraded model than advertised. It also runs canary style checks for hidden prompt injection and tests whether streaming responses work correctly. Results are labeled with a status like available, model possibly mismatched, suspected prompt injection, high risk, or pending retest, rather than jumping straight to a firm verdict from a single failed check. Users can submit new sites or report problems either directly on the website or through GitHub issues, and reviewers approve submissions before they go live. Contributors who provide detection keys for the checking system are listed on a public honor board showing their contribution counts and token usage, though the keys themselves are never shown or stored in this repository. The README includes a clear disclaimer: external sites change over time, users should judge risk themselves, and should never submit sensitive account details, verification codes, payment information, or private API keys through any listed site.
A docs-only landing page for a Chinese website that lists free and discounted AI API relay sites, sorted into risk zones with a model trust detection system.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly general.
This repo across BitVibe Labs
double-check against the repo, no cap.