Skip to content

Unable to switch reasoning effort and image input in Codex

Quick fix

v3.16.5 resets model catalog capabilities to defaults; editing the catalog is temporary and may be overwritten after switching.

Error output
无法切换推理强度和图片输入
English translation

Unable to switch reasoning effort and image input.

On Windows, the reporter connected gpt5.5 through cc-switch in the Codex desktop app and saw only the high reasoning effort level, with no image input support. Comments also noted that v3.16.5 writes only none and high reasoning levels to cc-switch-model-catalog.json; another report said edited settings were automatically reverted.

The original post says manually editing cc-switch-model-catalog.json can restore xhigh reasoning effort and image input, but the change returns to the default after switching. A user also reported deleting and rebuilding the Codex configuration, but did not confirm a stable resolution, so it is not a verified fix.

  1. As a temporary workaround, manually edit cc-switch-model-catalog.json as described in the original post: add an entry with effort set to xhigh in supported_reasoning_levels; set input_modalities to ["text", "image"] and enable supports_image_detail_original. The original post reports that switching automatically restores the defaults, so confirm whether these capabilities remain after editing.

  2. The original post does not provide a verified persistent solution. If the manual edit is overwritten, continue following updates in the source Issue.

ToolCodex
Versionv3.16.5(报告版本,未确认已修复)
PlatformsWindows

(No FAQ yet)

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.