This roundup covers the most common pain points when using ChatGPT: failed message sending, inability to switch models, repeated login verification, account security alerts, and more. The content is written as much as possible in the order of “check yourself first, then troubleshoot,” so you can quickly pinpoint the issue and get back to normal use.
Messages won’t send or keep spinning: rule out network and browser issues first
When ChatGPT shows “Send failed” or “keeps loading,” it’s usually related to the network path or browser cache. First switch networks (Wi‑Fi/cellular) or change browsers, and disable extensions that may block scripts (ad blockers, privacy plugins).
If only a specific page is acting up, try clearing this site’s cache and cookies and then log in again; if that still doesn’t work, open ChatGPT in an incognito/private window. Corporate or public networks often restrict WebSocket connections; switching to a personal network is usually more reliable.
Model switching unavailable: check permissions, service status, and the conversation itself
If model options in ChatGPT are greyed out or won’t switch, common causes include temporary server-side congestion, your account not having access, or the current conversation having enabled incompatible capabilities. Refresh the page and test by starting a new chat, rather than repeatedly switching within an old thread.
If only one particular model is unavailable, go to the settings page to confirm your account status is normal and whether you need to log in again. When the platform is experiencing an outage, the most effective approach is to try again at a different time, rather than repeatedly clicking to switch and triggering more frequent verification.


