iOS regression: Projects sidebar list removed and intelligence selection broken in Project chats

What issue are you seeing?

Summary

The latest Codex experience on iOS introduces two significant regressions affecting Projects:

1. The expandable Projects list has been removed from the sidebar.
2. Intelligence can no longer be selected normally when starting a chat inside a Project, and the hidden long-press selection does not reliably route the conversation to the selected intelligence.

Regression 1: Expandable Projects list removed from sidebar

Previously, the Projects section in the sidebar could be toggled open to display an expanded list of projects directly.

That option appears to have been removed. There is no longer a way to keep the project list expanded in the sidebar, making project navigation slower and less convenient.

Expected behaviour

Users should be able to expand or collapse the Projects list in the sidebar, as they could previously.

Actual behaviour

The Projects toggle and expanded sidebar list are no longer available.

Regression 2: Intelligence cannot be selected normally inside a Project

Previously, when entering or starting a chat inside a Project, the control at the top allowed the user to select the Project and choose the desired intelligence level.

That control now appears to switch between Work and Chat instead. As a result, there is no visible or obvious way to select or change the intelligence level when starting a Project chat.

When starting a new chat inside a Project, the user should be able to clearly select the desired intelligence level before sending the first message.

Actual behaviour

The normal intelligence selector is no longer available in the Project chat interface. The former control now switches between Work and Chat instead of allowing intelligence selection.

Regression 3: Long-pressing Send and selecting Pro still results in Medium

A hidden intelligence menu can be accessed by long-pressing the Send button. However, selecting Pro there does not appear to route the resulting conversation correctly.

Expected behaviour

Selecting Pro before sending should route the conversation to Pro and preserve that selection for the resulting thread and its follow-up messages.

Actual behaviour

Although Pro is selected through the long-press menu, the resulting conversation shows Medium for follow-up responses. This indicates that the selected intelligence is either not being applied, not being persisted, or the conversation is being routed to the wrong model.

What steps can reproduce the bug?

1. Open a Project in the ChatGPT iOS app.
2. Start a new chat inside the Project.
3. Observe that there is no visible intelligence selector.
4. Long-press the Send button.
5. Select Pro.
6. Send the message.
7. Continue the conversation or inspect the intelligence shown for subsequent replies.
8. Observe that the conversation shows Medium, rather than Pro.

What is the expected behavior?

_No response_

Additional information

Why this matters

This creates both a discoverability problem and a functional routing problem:

• Users cannot clearly choose intelligence when starting a Project chat.
• The only available workaround is hidden behind a long press.
• Even when Pro is selected through that hidden menu, the conversation may still run at Medium.
• Users therefore cannot trust that Project chats are using the intelligence level they selected.
• Removing the expandable Projects list also makes navigation between projects less efficient.

Environment

• Platform: iOS
• Surface: Codex within the ChatGPT iOS app
• App version: [add version]
• Subscription: [add plan]