# All API Hub: managing many AI relay accounts from one browser extension

> A browser extension that treats scattered relay station credentials as assets to be tracked, compared, checked in and exported, plus a manager for the gateways behind them.

**qixing-jk/all-api-hub** — All-in-one New-API/Sub2API account hub: balance/usage dashboard, auto check-in, one-click keys, price comparison, health checks, plus advanced channel management.

- Repository: https://github.com/qixing-jk/all-api-hub
- Website: https://all-api-hub.qixing1217.top/
- Stars: 4,917 · Forks: 330
- Language: TypeScript
- License: AGPL-3.0
- Published: 2026-10-07 · Updated: 2026-10-07 · Language: en
- Canonical page: https://hysenlabs.com/projects/qixing-jk-all-api-hub

## The problem statement is about scattered credentials

The README opens with a description of the situation rather than the software. AI relay sites are characterised as a marketplace for AI credits offering low-cost or even free access to models such as ChatGPT, Claude and GPT Image. The problem named is what happens once you hold more than one account.

Four complaints are listed. Assets are scattered, with balances and usage spread across separate sites. Pricing is messy, with different billing ratios making the best deal hard to spot. Daily perks get missed because manual check-ins are easy to forget. And setup work is annoying, because API details must be copied into tools such as cc-switch and Cherry Studio again and again.

Each of those maps to a feature group, which is a sign the tool was designed around real friction rather than a feature list. The framing phrase used is that this is your AI asset manager: add your site URLs and the extension handles the rest.

That framing also sets the ceiling. Every capability depends on the relay sites exposing an interface the extension can talk to, which is a moving target maintained by many independent operators rather than one company. The topics list names the ones it targets, including new-api, one-api, sub2api, openrouter, voapi, cherrystudio, cc-switch, codex, claude, claude-code and cliproxyapi, which is the vocabulary of this corner of the ecosystem.

## Balances, price comparison and check-ins

The account dashboard is the core. It manages multiple relay accounts in one place and reviews balances, usage and account status without signing in to each site. Alongside it sits an API credential library with an explicit independence property: no site account is required, so base URLs and API keys shared by others or collected over time can be brought together, then checked for balance, listed models, connection tests, or export.

Usage and trend analytics track balance changes with charts broken down by site, account and model, which answers the question of where the credits actually went rather than just what remains.

Cross-site model price comparison compares the effective price of the same model across sites. The README's framing is finding a better-value endpoint, and the word effective matters here, since different sites apply different billing ratios to the same underlying model and a raw per-token number would mislead.

Multi-site auto check-in handles the daily-perk complaint: check in with one click or on a schedule, collect rewards automatically, and skip daily logins.

A separate group handles announcements and task alerts. Site announcements are collected and displayed in one place with alerts for maintenance, model and pricing changes, and after an automated check-in, WebDAV sync or model sync completes, a result notification fires through the browser or a configured notification method, so failures surface rather than passing silently.

## Capture and export, which is where the time is saved

Quick web capture is the feature that removes the copying the README complains about. You find a base URL and an API key on a page, test them immediately, and save them to the credential library with less copying and tab switching.

One-click client export then sends credentials to named targets: CherryStudio, CC Switch, Claude Code Router and Kilo Code, with a documentation page listing the full set of supported tools. This is the part of the extension that saves the most time, since the alternative is reading a key off a screen and pasting it into each client.

Verification comes next. API and model verification checks connectivity and model availability in one click, so you can see whether a key works and a model can be called before you configure anything. CLI integration testing goes further, confirming that common command-line tools can use the target API before you spend time troubleshooting a full setup.

The logic is sound: a broken key discovered at the export step costs minutes, and the same key discovered during a model request inside a client costs an afternoon of debugging that has nothing to do with the client.

## The second half is gateway management, which is a different job

Past the client features the extension grows a second capability: managing self-hosted AI gateways directly, without opening each admin panel. Named targets are CLIProxyAPI, New API, Sub2API, AxonHub, Claude Code Hub, Octopus, Veloera and DoneHub.

From that, two more capabilities follow. Gateway channels can be created quickly by turning saved site accounts or credentials from the library into channels, so models can be called through the gateway with switching between channels as needed. And on supported sites, channel model lists can be synced manually or on a schedule as upstream models change, with custom redirects defined so clients can use the model names you prefer.

This is worth separating from the rest, because it changes the risk and the scale of the project. Managing a list of accounts you hold credentials for is one thing. Managing gateway channels, model lists and redirect rules is administering a serving tier that other tools and possibly other people depend on.

It is also where the churn shows. The v3.62.0 release notes are almost entirely account management: grouping and refining account sorting, group sorting controls with a settings shortcut, and reviewing suspected duplicates while reducing check-in interruptions. Two more recent changes follow the same shape, separating filtering from the pricing pipeline and preserving native routing defaults during import and migration. Account management is where the recent attention has gone.

## Storage, encryption and what local-first means here

The security section makes a claim that deserves reading closely rather than skimming. API keys, accounts and settings stay in your browser by default and are written to WebDAV only when you enable backup or sync.

That is the right default for the data involved. An extension holding dozens of third-party API keys has no business sending them anywhere by default, and a local-first design means the credentials are not a server-side honeypot. The trade-off is that clearing browser storage or switching machines loses everything, which is exactly what the sync feature addresses.

Encrypted WebDAV auto-sync is the answer: enable encryption and scheduled auto-sync to keep data synchronised across devices and pick up where you left off after changing computers. The encryption is the load-bearing word. Unencrypted WebDAV would put API keys for every relay account you use into a file on a third-party server, so whether that option is on by default is the detail worth checking in the documentation rather than assuming.

The repository reports AGPL-3.0 as the license. That is the most consequential fact on the page for anyone who intends to build on this code rather than install it, since AGPL's network use clause reaches software offered over a network, not only distributed binaries. For personal use as an extension it does not affect you.

## Distribution, versioning and the release cadence

All API Hub ships as a browser extension on three stores, with version badges for Chrome, Edge and Firefox. The README states a preference plainly: for most users the store build is recommended, because it is easy to install and supports automatic updates.

Versioning is unusually granular for a browser extension. v3.62.0 shipped on 2026-09-13, and a rolling `nightly` prerelease followed on 2026-03-15 with a build that names its package version, source commit, build time, source branch and trigger, plus a list of the commits since the previous version. A rolling tag like that is a CI convenience, and reading it alongside the versioned releases is the only way to tell a deliberate release from a snapshot.

The channels section is where the README is cut off mid-table, so the alternative install methods, if any, are not visible here. The documentation site covers getting started, supported sites, integrations and a changelog, and there are also Chinese and Japanese README siblings alongside the English one.

The project also links a substantial set of community spaces: a Linux.do thread, a WeChat group, a QQ group, a Discord community and a Telegram group, several of them explicitly multilingual. With 4,860 stars, 326 forks and 65 open issues, the community infrastructure is proportionate to the audience rather than bolted on.

The last push to `main` was 2026-09-14, the day after v3.62.0.

## Conclusion

The organising idea here is that a pile of relay station credentials is an asset portfolio, and the extension is a ledger for it: balances in one view, prices comparable across sites, daily check-ins automated, and keys pushed into whatever client you use rather than retyped. The local-first storage choice matters most for something holding this many secrets, and the encrypted WebDAV sync is the answer to the obvious objection about keeping API keys in a browser. Two things to weigh first. AGPL-3.0 is a real constraint if you build on the code rather than just install it, and the project also spans into managing your own self-hosted gateways, which is a second and larger job than the account dashboard. Install the store build, add one site and check the credential library flow before adding twenty, since understanding how credentials are stored is easier with a single entry to look at.

## FAQ

### What is All API Hub?

A browser extension for managing accounts on AI relay sites, the marketplaces that sell access to models like ChatGPT and Claude at discounted rates. It consolidates balances and usage into one dashboard, compares model prices across sites, automates daily check-ins, stores credentials locally, and exports keys into clients such as CherryStudio and CC Switch.

### Does All API Hub send my API keys anywhere?

Not by default. The README states that API keys, accounts and settings stay in your browser and are written to WebDAV only when you enable backup or sync. If you turn that on, encrypted WebDAV auto-sync is the documented option, and the encryption is what keeps the keys from sitting readable on a third-party server.

### What relay sites does All API Hub support?

It is built around the New API and Sub2API family of relay software, and the repository topics name new-api, one-api, sub2api, openrouter and voapi among others. There is also a dedicated supported-sites page on the documentation site, which is the authoritative list, since compatibility depends on what each operator's deployment exposes.

### Can I use All API Hub to manage my own AI gateway?

Yes, that is the second half of the project. It manages CLIProxyAPI, New API, Sub2API, AxonHub, Claude Code Hub, Octopus, Veloera and DoneHub from inside the extension, including creating gateway channels from saved credentials, syncing channel model lists on a schedule, and defining custom model redirects for clients.

## Sources

- [License: AGPL-3.0](https://github.com/qixing-jk/all-api-hub/blob/main/LICENSE)
- [Project website](https://all-api-hub.qixing1217.top/)
- [qixing-jk/all-api-hub on GitHub](https://github.com/qixing-jk/all-api-hub)
- [README](https://github.com/qixing-jk/all-api-hub/blob/main/README.md)
- [Releases](https://github.com/qixing-jk/all-api-hub/releases)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/qixing-jk-all-api-hub