The ChatGPT Windows desktop application was working normally until approximately 3–4 days ago. It then began intermittently hanging. Closing and reopening the application temporarily resolved the issue. As of today, the application consistently hangs and is unusable.
The ChatGPT web application functions normally in all browsers. The issue is isolated to the Windows desktop application.
Environment
Operating System
Windows 11 Pro
Version 25H2
OS Build 26200.8737
ChatGPT Desktop
Version 1.2026.133
Installed as MSIX package:
OpenAI.ChatGPT-Desktop_1.2026.133.0_x64__2p2nqsd0c76g0
Security Software
OpenText Core Endpoint Protection 26.2
Centrally managed through Web Console
Cannot disable shields locally
Symptoms
Desktop application launches successfully.
Clicking Log in does nothing.
Typing any prompt causes the conversation pane to become blank.
Application remains running but becomes unresponsive.
The Chat GPT application for Windows having issuess
Mine no loading the conversations, even after reseting or reinstalling not working at all
But Codex is working fines
New conversations titles appears but won’t load it’s content
Mobile and Web working fine which discard account or back end problems
Same:
Windows 11 Pro 25H2
ChatGPT Desktop 1.2026.133.0 (Microsoft Store)
Browser and Android work normally.
Repair, Reset, uninstall, reinstall, and reboot did not help.
Upper-right Log In button does nothing.
Lower-left Log In closes the sidebar but does not open authentication.
Sign Up button also does nothing.
Entering a prompt while logged out results in a blank screen.
Codex desktop app launches normally.
I have Windows 10, and I’m experiencing the same issue with the ChatGPT Desktop app. I’ve tried everything… even Codex. It managed to fix the sign-in issue, but it only worked for a short while. The problem is clearly somewhere in the app itself.
That is a fair question, and it has come up before. Here is my answer more or less from more than a year ago.
Think of the advancement of generative AI like a subway system with both local and express trains. The local train makes many stops, which is like doing extensive testing at every stage. The express train bypasses many of those stops, which is closer to moving quickly with less visible testing before changes reach users.
If the goal is to reach the future faster, the express train gets there sooner. If the goal is to check every stop along the way, the local train is safer and more predictable, but slower.
This is a shift from what many of us are used to, and it can be frustrating. After riding the express train for the last few years, I have learned to guard against certain types of problems because some rough edges seem to be part of the current pace of development.
Some of my safeguards include using Git, along with a daemon that actively backs up my ./codex and ./claude directories.
So I am not disagreeing with you. I am just describing how I have adjusted to the current reality. And yes, a few years ago I could easily have seen myself asking the same question.
For a different analogy, look at rocket development over the last century. SpaceX changed expectations by moving faster, testing in public more often, and accepting visible failures as part of the development process. That approach can accelerate progress, but it also means users and observers see more of the rough edges along the way.
