DeepSeek outages and problems
User reports, last 24 hours
9 reports in total, counted in 15-minute windows
No problem reports came in for DeepSeek in the last hour, and at this time of day there are usually almost none. Over the last 24 hours there were 9 reports. Worldwide, most recent reports were about app (100%). The last spike in reports was 188 h ago and lasted 1 h 37 min. These are user reports from the US, where most people report DeepSeek problems.
Around the world right now: normal in 48 other countries. By country
- NormalClose to the usual level24 h · 9 reports
- ElevatedMore than 1.5 times the usual levelNone in this chart
- OutageMore than twice the usual level and far above the day's averageNone in this chart
- Usual levelWhat a usual day sees at that timeAlmost none at this time of day
In the past 24 hours, 9 user reports came in for DeepSeek. Counting only the 15-minute windows with a spike big enough to count as an outage, that comes to 100.00% uptime for the day.
The busiest 15-minute window had 3 reports, about 12 hours ago. That stays below our outage threshold, so it doesn't count as downtime.
DeepSeek in November 2026
- Problem reports in November
- 1,587
- Outage spikes
- 3
- Periods of elevated reports
- 0
Monthly report for November 2026
What went wrong
What people said was wrong in recent problem reports for DeepSeek worldwide.
Share of 4 recent reports that said what was wrong, from every country listed below.
DeepSeek reports around the world
In the last 24 hours, 41 problem reports about DeepSeek came in from 17 countries. Right now reports are at the usual level in every country.
- Outage now
- Problems now
- Normal
| Country | Now | Last hour | 24 hours |
|---|---|---|---|
| The US numbers above | Normal | 0 | 14 |
| Brazil | Normal | 0 | 3 |
| Saudi Arabia | Normal | 0 | 3 |
| India | Normal | 0 | 3 |
| Hong Kong | Normal | 0 | 3 |
| Poland | Normal | 0 | 2 |
| Peru | Normal | 0 | 2 |
| Colombia | Normal | 0 | 2 |
| The UK | Normal | 0 | 1 |
| Italy | Normal | 0 | 1 |
| Canada | Normal | 0 | 1 |
| The Netherlands | Normal | 0 | 1 |
Each country's status compares its reports with the usual level there. 37 more countries had 5 reports in the last 24 hours. Map outlines: Natural Earth.
DeepSeek problem spikes, last 30 days
In the last 30 days, DeepSeek had 4 outage spikes. The longest lasted 2 h 5 min, on 3 November.
- Sat 24 Oct: 0 reports
- Sun 25 Oct: 0 reports
- Mon 26 Oct: 1 report
- Tue 27 Oct: 0 reports
- Wed 28 Oct: 6 reports
- Thu 29 Oct: 17 reports
- Fri 30 Oct: 54 reports (outage spike that day)
- Sat 31 Oct: 21 reports
- Sun 1 Nov: 14 reports
- Mon 2 Nov: 17 reports
- Tue 3 Nov: 309 reports (outage spike that day)
- Wed 4 Nov: 28 reports
- Thu 5 Nov: 19 reports
- Fri 6 Nov: 16 reports
- Sat 7 Nov: 376 reports (outage spike that day)
- Sun 8 Nov: 11 reports
- Mon 9 Nov: 22 reports
- Tue 10 Nov: 17 reports
- Wed 11 Nov: 22 reports
- Thu 12 Nov: 30 reports
- Fri 13 Nov: 20 reports
- Sat 14 Nov: 357 reports (outage spike that day)
- Sun 15 Nov: 20 reports
- Mon 16 Nov: 41 reports
- Tue 17 Nov: 100 reports
- Wed 18 Nov: 46 reports
- Thu 19 Nov: 64 reports
- Fri 20 Nov: 38 reports
- Sat 21 Nov: 14 reports
- Sun 22 Nov: 6 reports
-
Outage
Lasted 1 h 37 min162peak reports
-
Outage
Lasted 1 h 24 min143peak reports
-
Outage
Lasted 2 h 5 min111peak reports
-
Outage
Fri 30 Oct, 09:17–09:47Lasted 30 min35peak reports
By country
In the last 26 days, most DeepSeek problem reports came from the US (1,525), Germany (283) and Brazil (213). The US had the most days at outage level: 3 of 26. The latest outage spike was on Saturday 14 November, in the US, Germany, Brazil and 2 more countries.
- The US3 outage days1,525
- Germany2 outage days283
- Brazil1 outage day213
- The UK1 outage day203
- Indonesia1 outage day197
- India1 outage day · 1 elevated186
- Hong Kong2 outage days179
- The Philippines2 outage days177
- Outage spike
- Elevated reports
- Normal
- No data yet
Each square is one day. It turns red when reports stayed at outage level for half an hour or more, and amber when they stayed above the usual level for an hour or more. The number on the right is the country's total since 28 October 2026, when country history starts. A wider gap marks a Monday. Point at a square, or tap it, for that day's numbers. Days run from midnight to midnight, Central European time.
A spike is a sharp rise in problem reports from users, kept exactly as this page showed it at the time. Monthly reports for all services
How to fix DeepSeek not working
DeepSeek is up on our end, so if it won't work for you, the problem is most likely on your side. Work through these in order:
Your browser tries to reach deepseek.com over your own connection, right now. Nothing is sent to us or stored.
Resend in a new chat
When a reply hangs or you get the message saying the server is busy and to try again later, wait a minute and send your prompt again in a fresh chat instead of regenerating inside the old thread. Copy the prompt first so you don't lose it. If DeepThink was switched on, turn it off for one test, because reasoning replies take longer to finish.
Turn off your VPN or proxy
Some VPN servers and office proxies get blocked or throttled, so the page loads but your messages never send. Switch the VPN off, or try mobile data instead of Wi-Fi. If it works there, look at your network.
Clear DeepSeek site data or app cache
In Chrome, click the icon to the left of the address bar, choose Site settings, then Delete data, and sign in again. On Android, open Settings > Apps > DeepSeek > Storage and tap Clear cache; on iPhone, delete the app and reinstall it from the App Store. Old cookies and an outdated app build are a common cause of login loops and blank chat screens.
Check your API key and balance
If your own code gets errors while the web chat works for you, confirm the request uses an active key from your account and that your balance has not run out, because the API rejects calls from an account with no credit. The help centre at https://api-docs.deepseek.com/ explains each error response. Compare your failure times with the official status page at https://status.deepseek.com/ before you rewrite anything.
While DeepSeek is down: what to use instead
- Web chat instead of the app When the mobile app hangs on a reply that never finishes or keeps showing a server busy message, open the chat in a desktop browser, because one sometimes loads while the other fails.
- OpenRouter If your code calls the API, you can point it at DeepSeek models served through OpenRouter until the official endpoint answers again.
- Ollama on your own machine Smaller distilled R1 models run offline through Ollama; they are slower and weaker than the hosted chat, but they don't depend on anyone's servers.
- ChatGPT Fine for coding questions and drafts.
Fix DeepSeek on Windows, Mac, Android and iPhone
DeepSeek on Windows
- Fix
- Clear your browser cache and cookies, or temporarily disable third-party antivirus firewalls that might block the DeepSeek website or API. If you are developing with DeepSeek, verify that your Windows network proxy settings are not interfering with the connection.
- Check
- Try opening the DeepSeek website in a private browsing window to see if it loads. If you receive persistent gateway timeout errors, DeepSeek's servers are likely down.
ping deepseek.comTest whether DeepSeek's servers replynslookup deepseek.comCheck that DeepSeek's DNS resolvestracert deepseek.comTrace where the connection dropsipconfig /flushdnsClear the DNS cache, then reloadDeepSeek on Mac
- Fix
- Clear your Safari or Chrome browser data and disable any active VPNs or network extensions that could prevent DeepSeek from loading. Ensure your macOS system time is synced correctly, as SSL handshake errors can block access to DeepSeek.
- Check
- Check if you can reach the DeepSeek web interface using a different network, like a mobile hotspot. If the page fails to load across multiple browsers on your Mac, there is likely a wider DeepSeek outage.
ping -c 4 deepseek.comTest whether DeepSeek's servers replynslookup deepseek.comCheck that DeepSeek's DNS resolvestraceroute deepseek.comTrace where the connection dropssudo dscacheutil -flushcache; sudo killall -HUP mDNSResponderFlush the macOS DNS cache, then reloadDeepSeek on Android
- Fix
- Force stop the DeepSeek app in your Android settings, clear the app cache, and check the Google Play Store for updates. You can also try toggling Wi-Fi off to see if DeepSeek loads over mobile data.
- Check
- Open the DeepSeek app and look for 'server busy' or connection timeout alerts. If other network-dependent apps work fine, the issue is likely on DeepSeek's end.
DeepSeek on iPhone and iPad
- Fix
- Delete and reinstall the DeepSeek app from the App Store, or reset your network settings if you encounter persistent connection errors. Ensure that cellular data permissions are enabled for DeepSeek in your iOS settings.
- Check
- Check if the DeepSeek app presents a blank screen or fails to generate responses while your internet connection is otherwise active. You can also check the App Store for recent reviews indicating a shared DeepSeek outage.
Problems at the same time as DeepSeek
In the last 30 days, all 4 of DeepSeek's spikes in problem reports overlapped a spike at another service.
When several services get problem reports at once, something they share, such as a cloud host or a network, can be the real cause. Matching times alone do not prove it.
Most often at the same time
- Archive of Our Own4 times
- Lloyds Bank3 times
- Telkom3 times
- Workday2 times
- IRCTC2 times
- DeGiro2 times
How many of DeepSeek's spikes overlapped one of theirs.
During the latest spike
DeepSeek: Sat 14 Nov, 17:39 to 19:16
- Microsoft Outlook, outage,12 Nov 17:42 to 21 Nov 13:37
- Microsoft 365, outage,12 Nov 17:59 to 19 Nov 10:34
- Microsoft Store, outage,12 Nov 18:27 to 15 Nov 07:52
- 911, outage,12 Nov 23:04 to 15 Nov 07:20
- Workday, outage,13 Nov 19:08 to 15 Nov 07:35
- Gateway Fiber, outage,13 Nov 22:00 to 15 Nov 07:26
- IRCTC, outage,03:47 to 20:58
- 1&1, outage,06:18 to 17:56, 18:44 to 19:07
128 more services. Red dot: outage. Amber dot: elevated reports.
A spike is a sharp rise in problem reports from users, as the live pages showed it. Times are Central European time.
When people report DeepSeek problems
The busiest hour for reports about DeepSeek is 08:00 to 09:00 Central European time, and the busiest day is Tuesday.
By hour of the day
- 00:00 to 01:00: 0.63 reports on an average day
- 01:00 to 02:00: 0.44 reports on an average day
- 02:00 to 03:00: 1 report on an average day
- 03:00 to 04:00: 0.98 reports on an average day
- 04:00 to 05:00: 1 report on an average day
- 05:00 to 06:00: 0.72 reports on an average day
- 06:00 to 07:00: 0.8 reports on an average day
- 07:00 to 08:00: 0.44 reports on an average day
- 08:00 to 09:00: 2.6 reports on an average day
- 09:00 to 10:00: 2.3 reports on an average day
- 10:00 to 11:00: 0.82 reports on an average day
- 11:00 to 12:00: 0.26 reports on an average day
- 12:00 to 13:00: 0.72 reports on an average day
- 13:00 to 14:00: 1 report on an average day
- 14:00 to 15:00: 0.62 reports on an average day
- 15:00 to 16:00: 0.45 reports on an average day
- 16:00 to 17:00: 0.62 reports on an average day
- 17:00 to 18:00: 1.2 reports on an average day
- 18:00 to 19:00: 0.63 reports on an average day
- 19:00 to 20:00: 0.69 reports on an average day
- 20:00 to 21:00: 0.63 reports on an average day
- 21:00 to 22:00: 0.57 reports on an average day
- 22:00 to 23:00: 0.56 reports on an average day
- 23:00 to 00:00: 0.69 reports on an average day
By day of the week
- Monday: 13 reports on an average Monday
- Tuesday: 31 reports on an average Tuesday
- Wednesday: 17 reports on an average Wednesday
- Thursday: 27 reports on an average Thursday
- Friday: 17 reports on an average Friday
- Saturday: 30 reports on an average Saturday
- Sunday: 7.8 reports on an average Sunday
Each bar is the average number of user reports in that hour, or on that day of the week, from 46 days of reports since 5 October 2026. The busiest bar is highlighted. Spikes in reports are left out, so one outage does not set the pattern.
What happens when DeepSeek goes down
DeepSeek is a Hangzhou-based AI lab, founded in 2023, whose chat assistant runs in the browser and in mobile apps, while its API gives developers access to the V3 and R1 models. When it goes down, you lose a coding or writing session partway through, and any app built on the API stops answering its own users.
Most outages start with the chat replying 'The server is busy. Please try again later.' even though sign-in and your past conversations still load. Replies may also stop mid-answer. Web search tends to break before plain chat does, often with a notice that the search service is temporarily unavailable, and DeepThink answers from the R1 model slow down or cut off well before the faster default mode gives out. Trouble is usually partial rather than regional, because the API and the web chat can fail on their own, so your API calls may hang while the site looks fine; the official status page lists the two separately.
DeepSeek IP address, servers and network
- IPv4 address
- 3.173.21.63
- IPv6
- Not offered
- Network
- AS16509 · Amazon.com, Inc., registered in the United States
- Served through
- Amazon CloudFront · our check was answered by its Frankfurt (FRA) data centre
- DNS hosted by
- its own name servers (dnsv4.com)
- HTTPS certificate
- Issued by Amazon, valid until 15 May 2027 (173 days)
- HTTP/3
- Not advertised
- Our check
- Answered in 34 ms
Checked from our server in Frankfurt, Germany, 2 h ago, for deepseek.com. A service behind a CDN answers from many data centres, usually the one nearest to you, so these addresses are where its traffic enters, not where its own servers stand. Visitors elsewhere may see different addresses.
Checked from 12 places around the world
All 12 locations got an answer from deepseek.com. Fastest from Frankfurt (16 ms), slowest from Singapore (225 ms). Every location connected to the same address, on Amazon.com, Inc. (AS16509).
| Location | Result | Time | Address reached |
|---|---|---|---|
| FrankfurtGermany | Up | 16 ms | 3.173.21.63 |
| ViennaAustria | Up | 39 ms | 3.173.21.63 |
| MilanItaly | Up | 21 ms | 3.173.21.63 |
| ParisFrance | Up | 25 ms | 3.173.21.63 |
| AmsterdamNetherlands | Up | 17 ms | 3.173.21.63 |
| MadridSpain | Up | 72 ms | 3.173.21.63 |
| WarsawPoland | Up | 48 ms | 3.173.21.63 |
| DallasUSA | Up | 17 ms | 3.173.21.63 |
| Los AngelesUSA | Up | 27 ms | 3.173.21.63 |
| São PauloBrazil | Up | 183 ms | 3.173.21.63 |
| SingaporeSingapore | Up | 225 ms | 3.173.21.63 |
| TokyoJapan | Up | 69 ms | 3.173.21.63 |
Checked with check-host.net from its servers in these cities 2 h ago, for deepseek.com. Time is how long each server took to load the page.
DeepSeek FAQ
Is DeepSeek down right now?
Why do I keep getting "The server is busy. Please try again later."?
Why does Search say it's unavailable when normal answers still work?
Why do my API calls fail when the chat website works fine?
Why isn't my verification code arriving, or why can't I sign up?
Service updates from @deepseek_ai
No posts about problems from @deepseek_ai in the last 14 days. The latest is from .
Contact DeepSeek support
- Official status pagestatus.deepseek.com
- Help centreapi-docs.deepseek.com
- Websitedeepseek.com
- X (Twitter)x.com/deepseek_ai
Tip: when DeepSeek is down, its official status page is the quickest way to confirm the cause and see when a fix is expected.
About DeepSeek
- Category
- AI / Chatbot
- Founded
- 2023
- Country
- China
How we track DeepSeek
The status on this page comes from user-submitted problem reports for DeepSeek, counted in 15-minute windows over the last 24 hours. Each window is compared with the normal level for DeepSeek, so a few complaints on a quiet day don't trigger anything, but a sudden jump well above normal does.
When reports climb well above normal, the status here changes to partial outage or major outage and the chart moves with it. This is separate from DeepSeek's official status page: user reports tend to show a problem early, while the official page confirms the cause. If the two disagree, check both.