ChatGPT is taking forever to reply, or it’s stopped halfway through an answer. Start with OpenAI’s status page, then try the fixes below to narrow down the problem and get the conversation moving again.
Check whether ChatGPT has a service incident
Open OpenAI’s status page and look under System status for ChatGPT. Check the feature you’re using, such as Conversations, Login, or Voice mode, and open any active incident affecting it.
Read the incident updates for details about the affected features and recovery. An incident affecting one feature doesn’t mean every conversation is slow, and aggregate availability doesn’t rule out an account-specific problem.
For an earlier slowdown, open History, use Filter components, and open the relevant incident. Compare its updates with the time you had trouble.
Select Subscribe to updates on the status or incident page and follow the subscription prompts to receive service updates.
Get a stuck response moving again
Try these actions in order, stopping when replies work normally.
- 1.Wait 30–60 seconds for the response to finish.
- 2.If it remains stuck and the controls are displayed, select Stop generating, then Regenerate.
- 3.Start a New chat and resend your prompt, especially if the existing conversation has many turns.
- 4.Refresh the webpage or restart the ChatGPT app. For a hard browser refresh, press Ctrl+Shift+R on Windows or Linux, or Cmd+Shift+R on Mac. In Safari on Mac, Cmd+Shift+R opens Reader instead, so choose View > Reload Page or press Cmd+R.
- 5.Sign out of ChatGPT, then sign back in and retry.
For everyday questions, reduce extra thinking
For a straightforward question, check whether you’ve selected a reasoning or Pro option. The adjustments below apply to accounts with these controls; availability depends on your plan and rollout.
On eligible paid accounts with GPT-5.6 Sol controls, open the model picker or Thinking slider and choose Instant.
When your request needs reasoning, select Medium on the Thinking slider instead of High or Extra High if less reasoning is sufficient.
For a quick question in Work, switch to Chat. Work is intended for longer tasks.
On desktop, open the top-left menu and select ChatGPT.
Choose Chat in the top toggle.
On mobile, choose Chat from the top dropdown.
Try a private browser window
Close unnecessary browser tabs to free local resources. This addresses local resource pressure, not server response time.
Close desktop programs you aren’t using.
In desktop Chrome, open More > New Incognito window, or press Ctrl+Shift+N on Windows or Linux, or Cmd+Shift+N on Mac.
Open ChatGPT in that window.
Sign in to your account.
Submit the same request.
To check which extensions run in Incognito, open More > Extensions > Manage extensions.
Open a suspected extension’s Details. Start with script blockers, content modifiers, and privacy extensions.
Turn off Allow in incognito if it is enabled.
Repeat the request in the Incognito window.
If the private session works better, turn off suspected extensions in your regular window through More > Extensions > Manage extensions.
Tip: Test the same request in another browser or on another device to help isolate a local problem.
Clear ChatGPT’s saved site data in Chrome
If your regular Chrome window still has trouble, clear ChatGPT’s site data and be ready to sign in again.
- 1.Open More > Settings > Privacy and security > Third-party cookies.
- 2.Select See all site data and permissions.
- 3.Search for ChatGPT, select the site’s Delete control, and confirm. Delete old chat.openai.com data too if it appears.
- 4.Open ChatGPT again, sign in if prompted, and retry your request.
Compare your connections
Run an internet speed test when replies stall or fail to load.
Retry ChatGPT on another network. On a phone, compare Wi-Fi with cellular data; cellular service isn’t automatically faster.
Temporarily disconnect your VPN or proxy.
Retry the request.
If a secure-DNS or security filter is implicated in the connection problem, temporarily disable that filter using its own controls. The controls depend on the product.
Test the request again.
For trouble confined to a work or school network, ask IT to check OpenAI’s domain allowlist and allow ChatGPT WebSocket connections to wss://ws.chatgpt.com over TCP port 443, including a review of handshake filtering, connection timeouts, and message-size restrictions.
Tip: Use the same request for each connection test so you have a direct comparison.
When the slowdown survives the usual fixes
If ChatGPT remains slow across browsers, devices, and networks, open OpenAI’s Help Center and select the bottom-right chat bubble to contact Support. The initial interaction is with a virtual assistant.
Describe the slowdown and how to reproduce it. Include timestamps and your time zone, account plan, device and browser, selected model, conversation URL, and the troubleshooting you’ve already tried.
For browser diagnostics, right-click the page and select Inspect.
Open Network.
In Chrome or Edge, enable Preserve log. In Firefox, open the Network Monitor’s actions menu and turn on Persist Logs.
Clear the network log.
Reproduce the slowdown.
In Chrome or Edge, select Export HAR (sanitized).... In Firefox, choose Save All as HAR from the Network Monitor’s actions menu.
Review the file for sensitive information before sharing it. The ChatGPT desktop app doesn’t provide HAR-capture developer tools.
If the initial Support exchange doesn’t resolve the issue, provide the reviewed HAR file with relevant console errors and timestamps so Support can investigate further.
Frequently Asked Questions
Will upgrading to ChatGPT Plus fix slow responses?
OpenAI lists faster responses and priority during busy periods as Plus benefits. An upgrade does not repair an outage, and Plus still has usage limits.
Is a usage-limit notice the same as ChatGPT being slow?
No. Read the allowance or reset notice, then choose another available model or wait until the displayed reset time. Switching models does not reset shared allowances.
Can I make ChatGPT Voice answer faster?
Ask ChatGPT to speak faster to change its speaking style. This does not reduce server latency. For interruptions, use headphones and reduce background noise, or exit Voice and continue in text.