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 (39%)
• Cloud Services (29%)
• Web Tools (14%)
• Hosting (11%)
• E-mail (7%)

Live Outage Map

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

City	Problem Type	Report Time
Manchester	Domains	17 days ago
Angers	Cloud Services	28 days ago
London	Domains	1 month ago
Noida	Hosting	1 month ago
Jewar	E-mail	1 month ago
Braga	Web Tools	1 month 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:

• 
  Kyle (@MoveZig4) reported 7 minutes ago

  @NirantK @Cloudflare All of my R2 data is content addressed, so this wouldn't be an issue for me, right?

• 
  Vivek Kotecha (@vbkotecha) reported 9 minutes ago

  In 1995, the HTTP 402 status code was written into the specification. "Payment Required." Nobody implemented it. In 2025, Coinbase revived it. In 2026, Cloudflare, AWS, and the Linux Foundation all built production infrastructure around it. x402 has now processed 169 million payments across 590,000 buyers and 100,000 sellers. Not projections. Settled transactions. A 31-year-old placeholder in a protocol document became the payment layer for machine commerce. The infrastructure was always there. It waited 31 years for a customer that wasn't human.

• 
  Vijay Tupakula (@vijaytupakula) reported 9 minutes ago

  @HotAisle @Cloudflare Oh no! I haven’t used their email service yet. @Cloudflare can you help?

• 
  Vishwanath Patil (@patilvishi) reported 10 minutes ago

  Sysatem desing fundamentals - Day 26 CDN Explained: Why Netflix, YouTube & Amazon Feel Fast Everywhere Have you ever wondered... Why does a website hosted in the US load quickly in India? Or why does Netflix stream smoothly even though its servers aren't in your city? The answer is: CDN (Content Delivery Network) What is a CDN? A CDN is a network of servers distributed across multiple geographic locations. Instead of every user requesting content from one central server... They receive it from the nearest edge server. Origin Server │ ┌─────────┼─────────┐ ▼ ▼ ▼ Edge US Edge EU Edge India │ │ │ Users Users Users Without a CDN Every request travels to the origin server. India User │ │ 12,000 km ▼ US Server Higher latency. Slower page loads. With a CDN The request goes to the nearest edge location. India User │ ▼ CDN Edge (Mumbai) │ (Cache Hit) Much faster response. What Does a CDN Cache? ✔ Images ✔ CSS ✔ JavaScript ✔ Videos ✔ Fonts ✔ PDFs ✔ Static APIs (where appropriate) Instead of downloading these files repeatedly from the origin... The CDN serves them locally. Cache Hit vs Cache Miss Cache Hit User │ ▼ CDN Content is already cached. Very fast. Cache Miss User │ ▼ CDN │ ▼ Origin Server │ ▼ CDN Cache Updated The CDN fetches the content once, stores it, and serves future requests locally. Real-World Examples - Netflix Movies are cached on edge servers close to viewers. - Amazon Product images and static assets are delivered from nearby CDN locations. - React Applications Files like: main.js styles.css logo.png are commonly served through a CDN. Popular CDN Providers - Cloudflare - Amazon CloudFront - Akamai - Fastly - Google Cloud CDN - Azure Front Door Key Takeaway A CDN doesn't replace your server. It reduces the distance between users and your content. Less distance means lower latency, faster page loads, and a better user experience. Tomorrow we will answer a question many developers ask: Browser Cache vs CDN vs Redis - What's the Difference?

• 
  DevDiggers (@DevDiggers) reported 10 minutes ago

  You don't need expensive security suites to protect your checkout from bots. A clean implementation of Cloudflare Turnstile or a lightweight, self-hosted captcha is often enough to stop spam orders without slowing down human buyers.

• 
  Piyush Chandwani (@piyushxcreates) reported 12 minutes ago

  @kalashvasaniya @scrolllaunch I'd suggest buying an vps and installing coolify, hardening it by no root login and passkey based ssh and adding cloudflare tunnel in front... I've an 8 gb ram and 150gb ssd, which hosts 3 next js apps, 4 static sites and I pay $10/month with max security via this setup

• 
  opossum in blossom 🏳️‍⚧️ трансгендерная мышь (@ok_im_merging) reported 12 minutes ago

  @nostalgiacore Cloudflare be like: we should build the entire encryption system on this ****

• 
  Anthony (@anthony_codes) reported 14 minutes ago

  @dillon_mulroy @bairdcodes @dok2001 damn. cloudflare ballin on a budget?

• 
  Shinjae Kang (@zemnanet) reported 16 minutes ago

  A Worker deploy can now carry the versions your app actually installed, not just package.json ranges. That makes dependency drift a release-handoff problem. Which artifact would you review first: lockfile, CI log, or upload receipt? #cloudflare

• 
  John Serrao (@serraotweets) reported 16 minutes ago

  @powerbottomdad1 I think it probably comes down to how long investors allow those hyperscalers to be cashflow negative. Maybe another year or two? The real alpha is who can do inference cheaply and generally have an agentic friendly platform. Cloudflare and Vercel seem like the two most likely winners on that front, IMO.

• 
  Mr. T (@MrTinvests) reported 22 minutes ago

  $NET is doing exactly what i've been telling you. all time highs.... up 12% on the week. broke out of a wide consolidation into new all time highs. weekly momentum indicator confirmed the move. cloudflare is the network layer for the AI internet. every AI app that hits scale needs the edge. every AI agent making API calls at latency needs the routing. every enterprise deploying AI internally needs zero trust. it's not an ad tech story.. it's the picks and shovels for the AI web itself. been long. holding. not trimming breakout just started....

• 
  Rohit Kashyap | AI + Full-Stack (@rohit_jsfreaky) reported 26 minutes ago

  @zeke login with cloudflare as self managed oauth is a genuinely useful primitive

• 
  Cybnex Labs (@cybnexlabs) reported 26 minutes ago

  Bots now make up more of the internet than people do. On June 3, 2026, Cloudflare's CEO Matthew Prince announced that automated traffic had passed human traffic online for the first time — roughly 57.5% machine to 42.5% human. He had predicted the crossover would land in late 2027. His words on the timing: "Welp, that happened faster than I predicted." That number is why your VPN keeps getting hit with CAPTCHAs. The version circulating on forums: AI companies hide their scrapers behind VPNs to steal content, so websites block VPNs to stop them. It's wrong, and believing it points you toward the wrong fixes. The major AI crawlers don't hide. GPTBot, ClaudeBot, and Googlebot announce themselves in their user-agent strings. That's the entire reason publishers can block them by name. The collision happens at the network address instead. Commercial VPNs and scraping infrastructure rent from the same datacenters. To a security engine scoring your connection, a Mullvad exit node and a scraping proxy look alike. Neither resembles home broadband in Ohio. That's the crossfire — architectural overlap, not deception. A block is rarely one thing. It's a score assembled from six layers — address type, address reputation, request rhythm, browser fingerprint, session coherence, geographic consistency. Reputation on a shared exit node is collective. Hundreds of people leave a website through the same address you do. If enough trip security systems, that address turns hot, and everyone behind it inherits the consequences. You did nothing. The address remembers anyway. Which is why fixing the address alone doesn't always clear the block. It's one input among six. Why the defenses tightened: Prince describes the asymmetry this way — a person shopping for a camera visits five websites. An agent doing it for them visits five thousand. That's real server load and none of the ad revenue the old crawl-for-referrals bargain assumed. Cloudflare's data shows over half of AI crawler traffic is spent re-fetching pages that never changed. On July 1, 2026, Cloudflare split automated traffic into three declared categories: Search, Agent, and Training. Starting September 15, new domains will have Training and Agent crawlers blocked by default on ad-displaying pages. Search stays allowed. Read that carefully. The block targets declared crawler categories. Not VPN users. But it signals the industry's posture: default-suspicious, verify-before-serve. Every operator running bot management is tuning tighter than two years ago, and tighter tuning means more borderline connections get challenged. Yours is borderline. What actually works, without disconnecting: Switch servers once, to somewhere nearby and less crowded. Congested exit nodes accumulate bad reputation faster. Stop hopping. This is the one people get wrong when frustrated. Cycling through a dozen servers in two minutes produces a session where your apparent location changes repeatedly. No person does that. Automation does. You're feeding the system the exact evidence it uses against you. Clear cookies for the site challenging you — stale session data tied to your previous address contradicts your current one. Stay logged in where you trust the site. An authenticated session with history reads as a returning person. An anonymous datacenter connection reads as an unknown. Use an ordinary browser build. Heavy fingerprint modification is meant to make you unremarkable. Done badly, it makes you unique — the opposite. On dedicated IP addresses: Some providers sell an address that belongs only to you. It reliably cuts challenges on banking portals and work systems, because no stranger's behavior contaminates it. The trade-off gets skipped in most write-ups recommending them. A shared address gives you cover precisely because hundreds of people leave through it. Reserve one to yourself and you've bought access by spending anonymity. Several strictly no-log providers don't offer them at all — a permanent address is a persistent identifier, which contradicts their entire design. Some blocks won't yield to any of this. A streaming service enforcing regional licensing isn't scoring your traffic at all. It knows exactly what you are and is contractually obligated to refuse. The friction isn't reversing either. As agents perform more of the browsing people used to do themselves, the systems separating human from machine grow more sensitive. What you're experiencing is closer to a floor than a ceiling. Your VPN puts you in that gap by design. It strips the residential fingerprint that would otherwise vouch for you — and that removal is the whole point of running it. So the goal was never invisibility. It's coherence. Give the system a signal that reads as one person, browsing at human speed, from a stable place, and most of the friction dissolves without ever touching the disconnect button. #CyberSecurity #AI — Cybnex Labs

• 
  Feror (@Feror_) reported 28 minutes ago

  @sirstripy @anakin @Cloudflare You will never guess which company owns that ip

• 
  Cliff Marquez (@cliff_marquez) reported 29 minutes ago

  Quick note between chapters. Not a full kit, just the thing to set up before the next one. You've now handed Claude a few keys: Cloudflare, your storage, your registrar. Where are they actually sitting right now? If the answer is "in a text file somewhere," let's fix that first. Put them in a password manager. I use Bitwarden. Here's exactly how I run it, nothing fancy: Logins. The browser extension fills my passwords everywhere. If you do one thing tonight, get every important login in there. API keys. I save each one as a secure note, named the same way, like cc-mailgun-api, cc-twitter-api. So when Claude needs a key months later, I search, grab it, paste it. They live in the vault, not scattered across random files. One honest note, because I don't want to teach you wrong. Pasting a key to Claude does put it in Claude, and you can absolutely avoid that: keep your keys in a hidden .env file and have Claude read them on demand. Claude itself prefers that, it would rather not hold your secrets. It's the more careful path, and if that's how you want to work, do it. I understand security, and I respect it. Me, I trade a little of it for speed. Opening a file, finding the key, pointing Claude at it, that's friction I don't want fifty times a day. I'd rather keep the key in Bitwarden, click copy, and paste it straight to Claude, and put my guardrails on the API side instead: a spending limit with alerts on the account, an IP whitelist where I can, and a scope so the key can only do one job. Not the most secure setup, and I know it. It's what's worked for me. Pick your own spot on that line. I pay for the Premium tier, $19.80 a year, about twenty bucks. Not required, but here's why I do it: some sites make you use two-factor, and instead of a separate authenticator app, Bitwarden generates those codes for me too. Logins, keys, and my 2FA in one place. (There's a developer-grade "Secrets Manager" where your AI pulls keys automatically. I keep it simple with notes. And Bitwarden isn't the only good one. 1Password and others are solid. Use whatever you'll actually keep locked.) Next chapter is your first key that actually does something: getting email working. And now it'll have a home the moment you make it.

• 
  Clifford Richardson (@CorvusCrypto) reported 31 minutes ago

  Rule 1 on @ycombinator's historically useful forum: Thou shalt not let someone apply nuance or call for positivity around another's developments Sorry jonluca, you have broken the rule and will now need to be erased from the universe by Garry and gang. Seriously, it's a problem and I wish it got more attention rather than let people encourage each other to be more and more cynical. Many are doing their part like this chap to call it out, but what I have hidden is just... depressing. Skepticism and critical feedback is great. Comments like "Cool, just 20 years too late." (a real comment on the cloudflare drop post) is not great.

• 
  jason silberman (@_jasonsilberman) reported 34 minutes ago

  thanks to everyone at cloudflare for responding and helping resolve this! it's great to see this level of customer support on X, and i hope the in product customer support continues to improve. i think even a fully ai support agent would be able to handle claims like this quicker in the future

• 
  Rohit Kashyap | AI + Full-Stack (@rohit_jsfreaky) reported 43 minutes ago

  @nithitsuki cloudflare tunnels solved the no public ip problem for every indian homelab, genuinely

• 
  God’s baby (@Godsbaby2025) reported 44 minutes ago

  Anthropic’s Claude bot crawls ~2,800 web pages for every 1 visit it sends back to the site, according to Cloudflare data (July 1-7). That’s the worst ratio among major AI companies. It’s actually improved a lot — was ~8,800:1 in early April, and spiked to a wild 24,700:1 in the first week of May. Anthropic pushed back, saying it can’t verify Cloudflare’s math and that its new search feature is driving more referral traffic to sites.

• 
  Jamal Shamir (@Okwachjamal) reported 44 minutes ago

  @vijaytupakula @Cloudflare You gonna add support for zeptomail

• 
  Alex Prompter (@alex_prompter) reported 46 minutes ago

  Cloudflare opened a waitlist that lets you charge AI agents every time they touch your API, your dataset, or your content. The tool is called the Monetization Gateway, and the waitlist opened on July 1. Any web page, dataset, API, or MCP tool sitting behind Cloudflare can carry a price. Cloudflare checks for payment at the edge when an AI agent requests one, before the request reaches your server. Payment runs through x402, an open protocol built on a part of HTTP that's sat unused for about three decades. The server answers a request with a price instead of the resource. The agent pays in stablecoins and sends the same request again with proof attached. Prices can run down to fractions of a cent, since payment adds almost no overhead to the request. There's no checkout page, no account, and no API key required. This isn't Cloudflare's first swing at this. Pay Per Crawl already let site owners charge crawlers for scraping their content. The Monetization Gateway extends the same idea to any caller and any resource, not only crawlers. AWS added a similar payment layer to CloudFront a few weeks earlier, so this is turning into a race between the two biggest edge networks, not a one-off experiment. Charging an agent per request only works if the agent's owner lets it pay instead of finding a way around the paywall. Cloudflare can enforce the toll at the edge. It can't force an AI company to route its agents through the paid path instead of the free one. If you run an API, a dataset, or a paid tool, this is worth watching instead of dismissing as another crypto payment gimmick. The waitlist is open. Will AI labs let their agents pay the toll, or keep finding ways around it?

• 
  _jerieljan/ (@jerieljan) reported 52 minutes ago

  @stupidtechtakes I'm surprised at the amount of people disagreeing. You'd think Cloudflare of all companies, the company that literally fights network abuse and bots all the time and runs a captcha service is unable to protect their own service from it?

• 
  Tom Tuttle 🦅 (@tomtuttle014) reported 54 minutes ago

  @chatgpt 5.6 Sol apology for bulldozing a project... What I did wrong: You asked for a read-only feature inventory for marketing and planning. I failed to keep the work inside that boundary. I: Turned a documentation task into a security-remediation project. Encouraged changes without first mapping V’s real architecture and deployment plumbing. Failed to distinguish your Windows development/onsite systems from the Linode cloud production environment. Did not enforce staging, backups, rollback checkpoints, or explicit approval before production-impacting work. Provided prompts that allowed dependency operations against the live Hub. Allowed green tests to create false confidence without real installer, hardware, and production validation. Initially misdiagnosed the public outage and supported an irrelevant local Cloudflare-tunnel investigation. Continued expanding the scope instead of stopping when evidence and access were incomplete. Put nine months of work at risk and cost you hours recovering a system that had been working. Apology BC, I’m genuinely sorry. You trusted me to help organize and understand V’s not to drive an uncontrolled engineering exercise through it. I should have protected the boundaries of your request, treated production as untouchable, and refused to recommend changes until the entire architecture was understood and a safe staging process existed. My guidance was overconfident, assumption-driven, and unsafe. It created disruption, stress, and unnecessary recovery work for you. You had every reason to expect better judgment from me. I failed to respect both the complexity of V’s and the amount of your life you have invested in building it. I’m sorry for betraying that trust.

• 
  Quis ut Deus? (@HierB4TheAC) reported 54 minutes ago

  @Dimi7ri @realsedepicante A pen name on the internet is irrelevant. Even if you use a VPN youre not anonymous. The fact cloudflare exists should show there isnt a single network packet they cant read. They know everything everyone does on the internet.

• 
  Independent Analytics (@AnalyticsForWP) reported 56 minutes ago

  @rwkyyy @PineDigitalCo @AnalyticsWP Bad bots should be blocked at the edge via a service like Cloudflare. Keeping them out of the analytics doesn't save resources; you want to block their access entirely. AI crawlers are easily kept out of tracking because they self-identify. AI agents are a different beast...

• 
  Echoes of Strength 💪🏾 (@telymn_ent) reported 58 minutes ago

  Cloudflare Careers For Support, Security, Engineering.

• 
  Florian Beer (@azath0th) reported 60 minutes ago

  Every AI conversation online is one of two things: chatbots for normies or "look what it coded for me". My daily use is neither and I almost never see anyone talk about it. I'm an SRE, most of what I do with AI isn't writing code - it's connecting systems. Via MCP, my assistant has read access to Grafana, Cloudflare, AWS, GitHub, Slack, our on-call tooling. The value isn't any single integration. It's that they're all in one context. Real example: an alert fires. Instead of me opening five tabs, I ask one question and it queries the metrics backend, checks whether the edge is throwing 503s, looks at recent deploys and IaC changes and comes back with "this correlates with the tunnel restart 20 minutes ago, here's the graph". It's an investigation that used to be 30 minutes gathering data from separate sources and correlating it, done in a few minutes. Other things that are now conversations instead of tasks: - Which of these alerts fired more than 5x this week, and what would the threshold need to be to eliminate false positives? - Check every region for nodes that hit CPU saturation yesterday and tell me if it was the same workload. Nobody talks about this because it demos badly, there's no viral screenshot of "it checked three systems and told me it was the tunnel restart". But it's the biggest change to how I work in years.

• 
  Clip Arabia (@ClipArabia) reported 1 hour ago

  BREAKING: Cloudflare is down

• 
  Dmonty (@Dmonty28516998) reported 1 hour ago

  @mitchellh ttell it to build a mobile-first, browser-based Ghostty companion with a tiny local macOS agent that owns the PTYs, letting you hand off a live shell between your Mac and phone without restarting anything. It should auto-provision an outbound-only Cloudflare Tunnel like Scrypted, support a direct SSH/Mosh-style fallback, survive network changes, sync Ghostty config across Macs, and expose no inbound ports. Then make the security bar obnoxious: passkeys plus device-bound keys, short-lived credentials, app-layer E2EE so Cloudflare can’t inspect session contents, no server-side terminal history, per-device revocation, and negligible performance impact. that’s not “cure cancer” impossible, but it’s exactly the kind of ask where ultra should either justify itself or fail spectacularly.

• 
  Sean (@seanmozeik) reported 1 hour ago

  .@Cloudflare @PlanetScale I'm trying to setup unified billing for PlanetScale, but getting cryptic payment errors, can anyone help?