Duplicate user message shown three times when sending a prompt once

Resolved 💬 5 comments Opened May 12, 2026 by ceilf6 Closed May 12, 2026
💡 Likely answer: A maintainer (etraut-openai, contributor) responded on this thread — see the highlighted reply below.

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

26.506.31421

What subscription do you have?

ChatGPT Plus

What platform is your computer?

Darwin 25.3.0 arm64 arm

What issue are you seeing?

<img width="2048" height="1409" alt="Image" src="https://github.com/user-attachments/assets/66b661ab-de46-4f8d-ad7c-7ad3b9aee3ac" />

After I submitted a prompt once in a Codex conversation, the same user message appeared three times in the conversation UI.

The duplicated prompt text was:

遇到和图像相关的用HTML展示,例如这里展示Newton法的中间过程

In the UI, the first two duplicated messages stayed in a "thinking" state, while the third one eventually completed and showed a response. This made it look like the same prompt had been submitted multiple times, even though I only intended to send it once.

There was no explicit error message shown. The issue appears to be either a duplicate submission, retry, or conversation state synchronization problem.

I only clicked the send button once.

What steps can reproduce the bug?

I do not have a deterministic minimal reproduction yet, but this is what happened:

  1. Open a Codex conversation.
  2. Enter the following prompt:

遇到和图像相关的用HTML展示,例如这里展示Newton法的中间过程

  1. Submit the prompt once.
  2. Observe the conversation UI.

Actual result:

  • The same user prompt appeared three times in the conversation.
  • The first two duplicated entries stayed in a "thinking" state.
  • The third entry eventually completed and produced a response.
  • No explicit error message was displayed.

Session ID: Not available
Token limit usage: Not available / not shown
Context window usage: Not available / not shown

What is the expected behavior?

When I submit a prompt once, the Codex conversation should show exactly one user message and start exactly one corresponding assistant/Codex run.

If a retry happens internally due to network or server issues, the UI should not display duplicated user messages as separate conversation entries.

Additional information

_No response_

View original on GitHub ↗

5 Comments

etraut-openai contributor · 3 months ago

Thanks for reporting, but there's not enough information here for us to investigate further. If this happens again, please use /feedback to upload the session details.

ceilf6 · 3 months ago

This happened again.

feedback ID:
019e15df-f133-7aa0-8816-20ce0a899f86

Screenshot:
<img width="1128" height="745" alt="Image" src="https://github.com/user-attachments/assets/93b4c83a-adb8-4a03-8391-60ae61477c46" />

My concern is that this is not only a visual/UI issue:

  1. It may consume extra tokens unexpectedly.
  2. It may pollute the conversation context with duplicated or unintended messages.
  3. It may affect later Codex runs because the model may see stale, duplicated, or system-generated conversation state.
  4. It makes it hard to know whether my next prompt will be based on the real conversation history or on the duplicated/extra state.

<img width="1128" height="745" alt="Image" src="https://github.com/user-attachments/assets/6408bf70-6c65-4aec-ba5c-efc7f4e12df6" />

tangyang9464 · 3 months ago

I also encountered this problem, and it has existed for a relatively long time. Sometimes the message is sent twice instead of three times. feeback id: 019e39b7-ddf1-7ee1-a9b4-65a819b4e381

yonilevine · 3 months ago

I'm encountering the same issue. Feedback ID: 019e3bc9-00a4-7960-aa8e-8a8cab91f156

gitacoco · 2 months ago

I'm encountering the same issue again and again. @etraut-openai