Cloudflare status: hosting issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Cloudflare reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Most Reported Problems

The following are the most recent problems reported by Cloudflare users through our website.

• Domains (42%)
• Cloud Services (30%)
• Hosting (16%)
• Web Tools (8%)
• E-mail (4%)

Live Outage Map

The most recent Cloudflare outage reports came from the following cities:

City	Problem Type	Report Time
Crisfield	Domains	23 hours ago
Nanaimo	Web Tools	2 days ago
New York City	Web Tools	2 days ago
Istanbul	Domains	5 days ago
Greater Noida	E-mail	8 days ago
Paris	Domains	9 days ago

Full Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

Cloudflare Issues Reports

Latest outage, problems and issue reports in social media:

• 
  3th1c_yuk1 (@3th1c_yuk1) reported 3 minutes ago

  @krishnsec Yes, exactly. I reported an issue, and after almost two weeks it was finally reviewed. But by the time the triager attempted to reproduce it, the target had already been moved behind Cloudflare, so the issue was no longer reproducible and the report was closed as N/A.

• 
  George Pu (@TheGeorgePu) reported 6 minutes ago

  Anthropic launched MCP a year ago. Every AI lab built on it. Today Anthropic acquired Stainless - The company whose SDKs made MCP usable. OpenAI was a customer. Google was a customer. Cloudflare was a customer. Hosted products are being shut down. Anthropic isn't competing on models anymore. It's buying the rails everyone else runs on.

• 
  eustace bagge (@elhejjioui) reported 9 minutes ago

  @aso_web3 I mainly rely on systemd and Docker restarts so services recover automatically if something unexpected happens and for sensitive apps I use monitoring and Cloudflare DNS failover to a secondary VM. Never needed it so far but it’s there just in case

• 
  AGK (Ejike) (@dom000_dev) reported 10 minutes ago

  @deveasyclick @bahdcoder I don't know of those but cloudflare got it all, ever experienced most users on Glo or airtel not being able access or login your app but works for other networks? Cloudflare solves this latency with their reverse proxies, aside that many more goodies

• 
  Ravi Joseph (@rjkarmayogi) reported 10 minutes ago

  Technical marketing has the same structure as... marketing. Early in the history of the company, Cloudflare founder Matthew Prince realized that the output of Cloudflare’s engineering team could produce unique, differentiated marketing assets that could beat standard marketing content. He found that this created “the most cost effective and effective marketing that you can possibly do.” A key principle at work in Cloudflare’s engineering content: technical content still follows the same rules as other content types. There’s a different avatar, and specialized domain knowledge. But the structure underneath stays the same as in other types of marketing content: -Avatar -Problem -Solution -Transformation Cloudflare’s technical post “Our billing pipeline was suddenly slow. The culprit was a hidden bottleneck in ClickHouse” has this structure: -Avatar: Engineer running critical billing infrastructure. -Problem: Pipeline suddenly slow, no obvious errors. -Solution: Here's how we found lock contention in ClickHouse's query planner and patched it upstream. -Transformation: Query duration drops, resulting in competence and trust. In other words, Cloudflare’s technical post displays the same problem-solution-transformation structure as in other types of marketing. This problem-solution marketing structure is invisible when it's done well. The reader learns something, and in the process they learn that the company behind the content knows what it's doing. For infrastructure and developer tools companies, proof of competence can be a go-to-market strategy.

• 
  MG Reports (@martinOlsen94M) reported 11 minutes ago

  🇺🇸 Cloudflare just cut 1,100 employees. 20% of staff. First mass layoff in 16 years. Same quarter: record revenue of $639.8M, up 34% YoY. CEO Matthew Prince says 100% of code is now reviewed by autonomous AI agents. Internal AI usage up 600% in three months. Revenue went up Headcount went down. The agentic era just sent its first receipt

• 
  Haz (@diegohaz) reported 23 minutes ago

  Not gonna lie, I love Astro, but the Astro/Cloudflare migration from v5 to v6 has been the most painful one I've ever gone through, even with help from Opus 4.7 and GPT 5.5. Am I the only one?

• 
  Fern Holguin @GDC2026 (@HolguinDev) reported 23 minutes ago

  @DanzoProduction @Pirat_Nation Cloudflare Tunnel shutting off isn’t normal behavior, that’s usually config, ISP, or traffic related. Jellyfin is open source, there is no centralized account system to selectively shut users off. You should troubleshoot your issues that's the price of selfhosted stuff here

• 
  Siarhei Kernoga (@Kernoga_Siarhei) reported 24 minutes ago

  @Cloudflare “Patch faster” was never the right answer. The fundamental problem is the lack of architectural boundaries. If an AI can chain small bugs into meaningful access, that means the system allowed tiny local failures to compose into global authority. The fix is not just faster remediation - it is bounded access, explicit confirmation, scoped grants, isolated execution, and evidence by design.

• 
  Kohi needs coffee (@koshitantankohi) reported 27 minutes ago

  @weltjoachim8 They seem to have added cloudflare and a captcha. That's probably why it was down for a while.

• 
  CDSL – Central Depository Services India Ltd (@CdslIndia) reported 27 minutes ago

  @SHIVCHATTERJEE Dear Valued Customer, thank you for writing to CDSL, please DM us the details of your public IP address and Cloudflare Ray ID, for us verify reason of blocked

• 
  Rhoma, a tragedy in 97 acts 🏒🔥🚒🧯🏒 (@Rhoma_Ash) reported 30 minutes ago

  Anyone else getting ao3 cloudflare errors?

• 
  Josh (@joshmanders) reported 31 minutes ago

  Cloudflare is just microservices as a service

• 
  Tom Zarebczan 🛡 (@TomZarebczan) reported 32 minutes ago

  @DasCanard @dok2001 Sign in with Cloudflare - use case would be some app or service that would also request permissions for certain things and it auto creates a token for you.

• 
  Franklyn (@tfranklyn_) reported 32 minutes ago

  Day 86: Been working on file upload functionality and needed a way to cancel midway, why am I just hearing abt AbortController in js? I'm annoyed because something similar exists in Dio flutter. Deciding between cloudflare r2 or supabase for now #100DaysOfCode

• 
  Furkan (@TheRealFurkan) reported 33 minutes ago

  @ssgeos Why not move it behind Cloudflare? it would fix all your issues

• 
  Today's Startup News (@todaystartupnws) reported 34 minutes ago

  Anthropic confirmed Monday it acquired Stainless for a reported $300M+. The New York startup generates SDKs for OpenAI, Google, Cloudflare, and Anthropic itself. All hosted products shutting down; existing customers keep full SDK rights. Alex Rattray joins Anthropic.

• 
  LukeYoungblood.eth 🛡️ (@LukeYoungblood) reported 37 minutes ago

  @curtjg1971 @Cloudflare We put in a limit request Monday but still no resolution. They have really good service otherwise, but I hope the people handling limit increases are eventually going to resolve it.

• 
  Raja Patnaik (@RajaPatnaik) reported 39 minutes ago

  The line that matters in the @Cloudflare write-up is "faster patching is the wrong reaction." That an AI found bugs across fifty real repos was roughly expected. We've known for a year that a frontier model plus a code execution environment can do basic offensive work. A defense plan built on "patch CVEs faster" has already lost. The attacker's discovery-to-exploit window is collapsing toward zero, and accelerating the defender queue from days to hours doesn't close that gap. The shape of the real answer is architecture that assumes the vulnerability will be found and contains its blast radius before any patch ships. Capability scoping at the call site. Per-request execution isolation. Network egress as default-deny. Memory-safe languages wherever migration is feasible. Vulnerabilities stay in the code, quarantined by design, with exploitation buying the attacker far less than it used to. Teams that ship that architecture over the next 12 months will look outrageously paranoid from the outside. They'll also be roughly the only ones whose stacks aren't getting steadily chewed by an agent loop running 24/7.

• 
  Luis Gonzá*** (@luisgonzaleznf) reported 39 minutes ago

  Vercel finally killed the surprise billing meme. Getting a $4k invoice because a botnet scraped your static site was pushing everyone to Cloudflare. Glad they actually shipped a fix

• 
  well.base.eth (@curtjg1971) reported 40 minutes ago

  @LukeYoungblood @Cloudflare since its coming from you, if I was cloudflare I'd take care of this problem for mr. youngblood. Luke is a genious folks if he says there's a flaw, there's definitely a flaw. go get em brother

• 
  Chen Avnery (@MindTheGapMTG) reported 41 minutes ago

  @Cloudflare The real finding here isn't that AI finds vulns faster. It's that your vulnerability surface now expands 24/7 because agents never sleep. Patching faster is a treadmill. The only honest architecture assumes you're already breached and limits blast radius by design.

• 
  Hank Smythe 🇨🇦 (@HSmythe10) reported 42 minutes ago

  @UK_Daniel_Card My guidance to any firm is to NOT be reactive to vulnerabilties via scan and fix model. Instead patch every 3 weeks with the latest security patches available for all systems in play. Simultaneously protect publicly exposed systems via Firewalls and WAF (like Cloudflare).

• 
  Bored Devops ☠️🛠⚙️ (@syntoythesis) reported 45 minutes ago

  @theo @r_marked The only question I want to know is which level you're service will sit. Conventional wisdom says right above AWS/Azute/GCP, but that feels too boring. "It's a CloudFlare+Vercel wrapper," feels like something you'd go for and I am here for it.

• 
  Sarthiii (agent/acc) (@SarthiB7) reported 53 minutes ago

  @AlexanderGreyDT I use cloudflare and railway. I love how dynamic routing and security with agent support works there.

• 
  Geethika Isuru (@geethika_isuru) reported 56 minutes ago

  Anthropic just paid $300M to kill a tool OpenAI was using. The tool is called Stainless. Most people have never heard of them. But OpenAI used them. Google used them. Cloudflare used them. Stainless built the SDKs that let developers connect to AI models. Every official Anthropic library. Every official OpenAI library. All built on the same quiet engine in New York. Then Anthropic walked in with a checkbook. The deal closed. The team got absorbed. And every hosted product is being wound down. No new signups. No new projects. The factory that powered half the AI industry is now exclusive to one company. OpenAI had already given up doing this work internally. They handed it to Stainless. Now that supplier belongs to their biggest rival. Existing customers keep what they built. But the doors are closed. This is Anthropic's fourth acquisition in six months. Bun. Vercept. Coefficient Bio. Now Stainless. The agent wars are not being won with better models. They are being won by quietly buying the plumbing. P.S. Connect if you're building something in AI. The next 18 months are going to be wild.

• 
  Nicolas Grenié (@picsoung) reported 57 minutes ago

  @Miguel07Code @HyperFrames_ Performances, I had issues rendering a 1min video with 15 videos clips it was timing out all the time needed to use Cloudflare workers to render

• 
  rook (@bheuthanasia) reported 57 minutes ago

  havent been able to each a bunch of websites since the power outage the other day. is this some vietnam bs again or should i blame cloudflare

• 
  Alex Kostyniuk (@kostyniuk00) reported 57 minutes ago

  @uwunetes That’s the funniest part, if you wanna play the role - play independently, not call for support in comments until you’re right and people support you. What was he even expecting with calling to @Cloudflare, bro, no one supports negativity and bad attitude

• 
  𝕌𝕊𝕄ℂ𝕊𝕠𝕠𝕟𝕖𝕣𝔽𝕒𝕟 🏈 🎸 🌊 (@usmcsoonerfan) reported 59 minutes ago

  @celestedtwt Been having larger Chrome issues this morning too. Safari was fine. In Chrome I had to dump DNS cache and clear Chrome settings, and reboot my Mac. Just dumb. This Cloudflare BS happening AFTER all that.