Request still uses the old session model, Codex
Quick fix
Set all relevant provider configurations in the failover queue to the required model; when continuing an old session, select the target model in the Codex…
Symptom
Section titled “Symptom”codex cli设置的请求gpt-5.2, 队列中供应商模型指定的gpt-5.4, 实际请求的还是5.2The request model configured in the codex cli is gpt-5.2, and the provider model specified in the queue is gpt-5.4, but the actual request still uses 5.2.
In CC Switch 3.15 Codex App old sessions, requests may route to the current provider while still carrying the model name from the old session. The maintainer explained that 4.0 rewrote local routing and added aggregation mode, which lets you select models from different providers directly in the Codex model selector. If the issue still occurs, provide the mode used, version, reproduction steps, and logs.
In another issue, the Codex CLI requested gpt-5.2, while the provider configuration in the failover queue was set to gpt-5.4; the actual request still used gpt-5.2. The maintainer recommended changing each relevant provider's individual model configuration to the required 5.4.
If you use the failover queue, set the model configurations of all relevant providers in the queue to the model you want to request.
If you continue an old session in the Codex App, try selecting the target provider model in the Codex model selector. If the issue still occurs on the latest version, include the direct, routing, or aggregation mode, version number, reproduction steps, and relevant logs in the issue.
Affected Versions
Section titled “Affected Versions”Source Issues
Section titled “Source Issues”This page is distilled from 2 real issues
(No FAQ yet)
Related problems
Section titled “Related problems”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.