openai / openai/codex

Chat gets stuck on “Thinking…” while the response is actually completed and appears after forking

Open
#46,636 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug session
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

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

GPT 26.915.31945

What subscription do you have?

pro x20

What platform is your computer?

Darwin 25.3.0 arm64 arm

What issue are you seeing?

During a conversation, Codex sometimes becomes stuck in the “Thinking…” state and never displays a response.

Waiting does not resolve it, and sending a follow-up message also receives no response in the original chat.

However, when I fork the affected chat into a new chat, the previously missing assistant responses are present in the forked conversation. This suggests that the responses were generated/stored successfully, but the original chat UI failed to render or update them and remained stuck on “Thinking…”.

The attached screenshots show:

  1. Original affected session: the conversation shows “Thinking…” / stopped state and does not display the assistant response.
  2. Forked session: the responses that were missing in the original session are visible, confirming that output existed.

No explicit error message was displayed.

What steps can reproduce the bug?
  1. Use Codex in an ongoing conversation.
  2. Send a message and wait for the assistant response.
  3. Occasionally, the chat remains on “Thinking…” indefinitely and no assistant response appears.
  4. Send another message; the original chat still does not respond.
  5. Fork the affected conversation into a new chat.
  6. In the forked chat, observe that the assistant responses missing from the original chat are already present.

This is intermittent and I do not yet have a minimal code reproduction.

Affected original session ID: 01a0b97d-f006-73d0-b3f0-4cdcc3e3a352
Forked session ID: 01a0b9a7-d368-7a33-ba38-7f9d5a5310a1

What is the expected behavior?

Once a response has been generated, it should appear in the original conversation without requiring a fork. The chat should leave the “Thinking…” state, render the response, and accept subsequent messages normally.

Additional information

Screenshots

Original affected session

The chat remains stuck and does not render the assistant response.

Image
Forked session

The previously missing assistant responses are visible after forking.

Image

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the intermittent stuck “Thinking…” state in an ongoing conversation, using the affected and forked session behavior described in the issue. Compare the original chat’s state and rendering with the fork, and consider the issue resolved when a generated response appears in the original chat, the state clears, and follow-up messages work normally.

Written by the indexing model from the issue text.

Assessment

Domain
desktop-dev
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.