ChatGPT Desktop Codex selects GPT-5.6 Sol but runtime rejects it; context compaction loops
Nobody has claimed this yet.
- 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)?
ChatGPT Desktop app — bundled Codex CLI 0.153.4
What subscription do you have?
ChatGPT Pro (signed in with a ChatGPT account)
What platform is your computer?
27.0.0 arm64 arm
What issue are you seeing?
Summary
In the ChatGPT Desktop app, an existing Codex chat shows “5.6 Sol, extra high” in the model selector.
However, the task returned this runtime error:
“The 'gpt-5.6-sol' model is not supported when using Codex with a ChatGPT account.”
The same chat then repeatedly attempted automatic context compaction and reconnecting (2/2), but failed with:
“Error running remote compact task: Codex ran out of room in the model's context window. Start a new thread or clear earlier history before retrying.”
Other Codex chats are working normally. This affected chat remains stuck.
I already submitted in-app feedback. I can provide the feedback/session ID privately if needed.
What steps can reproduce the bug?
In-app feedback was submitted; identifier available privately upon request.
What is the expected behavior?
The model selector should show only models that are actually available to the signed-in ChatGPT account.
If a chat exceeds the context window, Codex should provide a reliable recovery path without repeated failed compaction and reconnect attempts.
Additional information
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the issue in the ChatGPT Desktop app with bundled Codex CLI 0.153.4, checking the model selector and the context-compaction/reconnect behavior. Compare the selected model with the signed-in ChatGPT account's supported models; done means unavailable models are not shown and an over-limit chat has a reliable recovery path without repeated failed attempts.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- ai, desktop
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100