The Cursor AI (in ‘Auto’ mode) stops sending text after 3–5 queries. By the 6th query or later, it just shows a constant loading spinner with the message ‘taking longer than expected.’ I’ve already tried reinstalling the app and starting new chats. I am on the $20 Pro plan, and I’ve only used about 50% of my ‘Auto’ limits, with remaining capacity on my API usage as well. It’s currently impossible to use.
Hey Borys, thanks for flagging this.
We’re tracking this “Taking longer than expected” hang already, and we’ve seen similar reports across different plans, versions, and OSes. The eng team is aware.
A few things that have helped others in the meantime:
-
Disable HTTP/2 (then restart Cursor). Add to
settings.json(Ctrl+Shift+P → Preferences: Open User Settings (JSON)):"cursor.general.disableHttp2": true -
Reindex the project: File → Preferences → Cursor Settings → Indexing & Docs → Resync Index (or Delete and Reindex).
-
Start a fresh chat (Cmd+N / Ctrl+N) if it’s stuck — long sessions can sometimes trigger hangs.
-
Run network diagnostics: Cursor Settings → Network → Run Diagnostics and share a screenshot of the results.
To dig into your specific case, can you share:
-
Cursor version
-
OS
-
A Request ID the next time it freezes (chat menu → Copy Request ID)
@dny_ex and @javiermarcon - same request: if you can share Request IDs from a stuck run, that’s the most useful signal for us.
Related thread (more reports + workarounds): “Taking longer than expected”
Best,
Mohit
Version: 2.6.11
VSCode Version: 1.105.1
Commit: 8c95649f251a168cc4bb34c89531fae7db4bd990
Date: 2026-03-03T18:57:48.001Z
Build Type: Stable
Release Track: Default
Electron: 39.6.0
Chromium: 142.0.7444.265
Node.js: 22.22.0
V8: 14.2.231.22-electron.0
OS: Darwin x64 24.6.0
Hey, thanks for the Request ID + version.
I checked b9b6184e-5b9b-42ab-95f7-878a9e2794c1: the request completed successfully on the backend. The “hang” is due to the session size: that chat has 347 messages, so the accumulated context makes each new request much slower and can trigger the “taking longer than expected” spinner.
Best workaround: start new chats more often (Cmd+N). Fresh threads stay noticeably snappier.
A couple of notes:
-
You’re already on HTTP/1.1, so the HTTP/2 toggle won’t help here.
-
If you’re on a VPN, try disabling it to reduce latency.
Let me know if you still see this in short/new conversations.
1 Like
Hi! Since I’m on a Mac, the following steps were designed for it, but the general flow might work for you too. I deleted the app, then went into the registry and completely removed all connections and files created by it, including the Python libraries. After that, I emptied the trash, reinstalled the app from scratch, and then added all the necessary libraries.
Is this the first time you’ve encountered this issue?
Yes, but I lived with it for a month, just reinstalled it a couple of times. This was the last step before completely uninstalling it and switching to another system, because I couldn’t use the product.
@dny_ex I hear you. I went back and checked your recent activity on the backend, and you’re right that the workarounds haven’t been enough. Your recent requests are intermittently failing before any response is generated, so starting fresh chats alone won’t address the root issue.
This has now been escalated and is under active investigation with high priority. I can’t provide an ETA yet, but it’s no longer sitting idle.
A couple of things that may help in the meantime:
• Try a different model temporarily (for example, claude-4.6-sonnet or Auto mode). It may not resolve the issue completely, but it helps us determine whether the failures are model-specific.
• Try @Borys’s deep clean approach (post #14 above) — a full uninstall that removes all app data, cached files, and related libraries before reinstalling from scratch. Regular reinstalls haven’t helped most users, but this more thorough method did resolve it for at least one person in the thread.