Skip to content
CC Switch WikiUnofficial Community Wiki

Usage query must be a same-origin request - cc-switch

Quick fix

Switch to the generic or custom template in cc-switch and put the usage query URL straight into the request address, bypassing the same-origin restriction.

Symptom

Section titled “Symptom”
Error output
3.16.3升级后提示必须要同源请求
English translation

After upgrading to 3.16.3 it says the request must be same-origin

Cause

Section titled “Cause”

In cc-switch v3.16.3, some built-in templates added a validation requiring the API request address and the usage query address to be same-origin (the same domain). Since the API endpoint and the usage query endpoint of most third-party relay services are usually on different domains, usage queries fail after upgrading. The cluster also holds reports of enormous token consumption and spiking bills when using a relay reverse proxy (Codex or GPT models): cc-switch converts point-to-point protocol-to-protocol internally, which breaks the signature-based context caching Codex relies on. That conversion caching problem is still being addressed with new caching capability.

Fix

Section titled “Fix”

  1. On the cc-switch config page, do not use the restricted provider-specific template.

  2. Select 通用模板 (generic template) or 自定义模版 (custom template).

  3. In the request address field, enter your usage query request URL directly.

Affected Versions

Section titled “Affected Versions”
Toolcc-switch
Version>= 3.16.3
PlatformsWindows

Source Issues

Section titled “Source Issues”

This page is distilled from 4 real issues

FAQ

Section titled “FAQ”
Can claude code use codex subscription models?
It can — a gateway tool like tingly-box can adapt and convert it. I think newer versions of cc-switch have added some support too.
Is this fixed in the new version? I do not dare upgrade lightly, since if it stops working it is hard to roll back.
There is a way around it: pick the generic template and put the query request url in the request address. Or use the custom template, which has no such restriction.
Token consumption is enormous using a codex auth reverse proxy in claude code — one hello takes 24% of the context. Is token consumption normal for everyone after connecting codex?
That is because cc switch converts internally point-to-point, protocol to protocol, in a simple way. codex uses signature as the identifier for keeping context, so a straight conversion inevitably keeps losing the cache effect. The usage problem is still being worked on (it needs a new caching capability).
Section titled “Related problems”
Switching with cc-switch has no effect - CodexCodex does not hot-switch. Restart the Codex app completely after switching relay in cc-switch; a migration tool can recover lost sessions.4 issues mergedIt says 1.6 billion used — usage statistics spiking - CodexUpgrade cc-switch to v3.18.0 to fix the sub-agent history replay double-count bug; it rebuilds the polluted Codex usage data automatically.5 issues mergedTicking 1M for DeepSeek does nothing? cc-switch local proxy configWith the cc-switch local proxy on, do not enter [1m] and leave the main model empty. The [1m] suffix only works with the proxy off.4 issues merged

This is an unofficial community wiki with no affiliation to the cc-switch authors or the project itself. Its content is compiled from the project's public GitHub issues. This site distributes no software.