Skip to content
CC Switch WikiUnofficial Community Wiki

Model Does Not Support Image Input and Blocks Image Reading - Codex

Quick fix

Upgrade CC-Switch to v3.17.0 and resave the provider configuration. If an alias remains text-only, temporarily edit the catalog and report it.

Symptom

Section titled “Symptom”
Error output
在 Codex 中上传/粘贴图片时，Codex 会认为当前模型不支持图片输入，直接阻止图片读取。
English translation

When uploading or pasting an image in Codex, Codex determines that the current model does not support image input and blocks the image from being read.

Cause

Section titled “Cause”

When CC-Switch v3.16.5 generated `~/.codex/cc-switch-model-catalog.json`, it hardcoded the default `input_modalities` in the template as `["text"]`. Even if a custom Responses model actually supports images, Codex still blocks image input locally based on the model catalog. The maintainer stated that this was fixed in v3.17.0: only confirmed text-only models declare `["text"]`; other models declare `["text", "image"]` by default.

However, a user reported that using the alias `deepseek-v4-flash` in v3.19.2 still generated a configuration containing only `"text"`, indicating that the issue can still occur with aliases or unknown models. The maintainer did not confirm a final fix for this later variant in this issue.

Fix

Section titled “Fix”

  1. Upgrade CC-Switch to v3.17.0 or later.

  2. After upgrading, resave the Codex provider configuration so that CC-Switch regenerates the model catalog.

  3. Check the generated model entry. For a model that supports images, `input_modalities` should include `"text"` and `"image"`.

  4. If an alias or unknown model is still marked as text-only in the newer version, temporarily edit the model catalog manually. Note that CC-Switch will overwrite this change the next time it saves the configuration.

    ~/.codex/cc-switch-model-catalog.json
    "input_modalities": ["text"]
    "supports_image_detail_original": false
    "input_modalities": ["text", "image"]
    "supports_image_detail_original": true

  5. If the issue still occurs after v3.17.0, submit the CC-Switch version, model alias, actual requested model name, and generated catalog contents to the original issue for further investigation.

Affected Versions

Section titled “Affected Versions”
ToolCodex
Version3.16.5（v3.17.0 修复原问题；v3.19.2 有别名模型变体复现）
PlatformsWindowsmacOS

Source Issues

Section titled “Source Issues”

This page is distilled from 1 real issues

FAQ

Section titled “FAQ”
How should I configure it to avoid this warning?
Upgrade first and resave the provider configuration. If a model that supports images is still recorded as `["text"]`, temporarily change it to `["text", "image"]`. Saving the configuration again through CC-Switch will overwrite the manual change. There is no confirmed permanent configuration solution yet for the alias-model reproduction in v3.19.2.
Section titled “Related problems”
Codex desktop cannot see modelsSign in to official ChatGPT, enable “Keep official login when switching to third parties,” and enable local routing. Restart Codex.6 issues mergedCodex shows no model list and will not switch modelsTurn on keeping the official login when switching to a third party in cc-switch, and make sure you signed in to official ChatGPT in Codex once.6 issues mergedChanged files cannot show diff below. CodexDo not automatically fetch 火山方舟 Coding Plan models. Manually use deepseek-v4-flash, then restart Codex and CC Switch.1 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.