Read Aloud takes ~40 seconds to start in Codex Windows app

What version of the Codex App are you using (From “About Codex” dialog)?

Codex version: 26.818.2441.0

What subscription do you have?

ChatGPT Plus

What platform is your computer?

Microsoft Windows 11 Pro, Version 25H2, OS Build 26200.9168, 64-bit (x64)

What issue are you seeing?

The Read Aloud feature in the Codex Windows app takes an unusually long time to start playing audio.

When I click the speaker / Read Aloud button on a response, it takes approximately 40 seconds before the audio begins.

I tested the same response on the same computer and internet connection using ChatGPT on the web. ChatGPT Web begins reading the response in approximately 4–5 seconds, while the Codex app takes around 40 seconds.

This delay makes the Read Aloud feature impractical to use in Codex and often causes me to switch back to ChatGPT Web instead.

What steps can reproduce the bug?

Open the Codex app on Windows.
Open a conversation containing a normal text response.
Click the speaker / Read Aloud button below the response.
Wait for the audio to begin.
The audio takes approximately 40 seconds to start.
Open the same response in ChatGPT Web on the same computer and internet connection.
Click Read Aloud.
Audio begins in approximately 4–5 seconds.

What is the expected behavior?

Read Aloud should begin playing within a few seconds, similar to ChatGPT Web.

Ideally, the Codex app should start audio playback in approximately 4–5 seconds or faster instead of requiring around 40 seconds.

Additional information

This comparison was performed using the same computer, the same internet connection, and the same response content.

The issue appears specifically related to the Codex app's Read Aloud startup latency, because ChatGPT Web starts playback much faster under the same conditions.

I use Read Aloud frequently, so this latency has a significant impact on my daily use of the Codex app.

Reproduce Read Aloud in the Windows app using Codex 26.818.2441.0 on Windows 11, then compare startup timing with the same response in ChatGPT Web. Trace the Read Aloud playback startup path and identify why the app takes about 40 seconds; done means playback starts within a few seconds, comparable to the reported 4–5 seconds on the web.