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 (33%)
• Cloud Services (30%)
• Web Tools (15%)
• Hosting (15%)
• E-mail (7%)

Live Outage Map

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

City	Problem Type	Report Time
Paris	Cloud Services	8 days ago
New York City	Hosting	11 days ago
Manchester	Domains	1 month ago
Angers	Cloud Services	1 month ago
London	Domains	1 month ago
Noida	Hosting	2 months 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:

• 
  Brian Cardarella (@bcardarella) reported 6 minutes ago

  @mikehostetler I wouldn't build anything around what @flydotio offers, they just throw **** against the wall. Look at what @dillon_mulroy and his group over at Cloudflare are offering.

• 
  Sudotech Limited (@SudotechLimited) reported 7 minutes ago

  X thought we were using a VPN. Here's what happened. MTN Nigeria uses a proxy. If you're on MTN, your traffic is automatically proxied. How we know: Logging into our Cloudflare account on Airtel always fails. On MTN, we can access Cloudflare and a few soft-banned service in Nigeria.

• 
  Vikas(Vik) Malpani| AI for US Real Estate (@vikasmalpani) reported 8 minutes ago

  @dhh @ashirsc @Cloudflare We run the same loop for closing exceptions. An agent catches the missing doc or mismatched figure, drafts the write-up with the evidence, and files it for a human to clear. The win isn't fewer errors, it's that every exception arrives already triaged.

• 
  Muhammad Tahir (@Ta56980T) reported 9 minutes ago

  @Cloudflare we have to stop solving problems for sake of problems, but focus on how to Never let those problems even happen (find out of box solution that why it even happens? how to make such technologies which are resilient to such attacks by their architecture.

• 
  Abdul Rauf (@armujahid) reported 9 minutes ago

  @karachism @argamingpk1 Will check. My setup is -> pihole (custom block list, dns cache) -> dnscrypt -> cloudflare zero dns. But yeah, noticed random issues on vanilla network without any custom setup.

• 
  nik skld (@nikskld) reported 15 minutes ago

  ANTHROPIC SHIPPED A CRON SCHEDULER FOR AI AGENTS AND ALMOST NOBODY COVERED IT. Every thread you’ve seen about Claude Managed Agents is still recycling the April 8 launch post. Three months later the product barely resembles that announcement. What actually shipped since: •Scheduled deployments (June 9). Give an agent a cron schedule. Every time it fires, the agent opens a fresh session and completes the task. No scheduler to build, no scheduler to host. •Environment variables in vaults. Register an API key with the domains it’s allowed to reach. The sandbox only holds a placeholder. The real key gets attached at the network boundary, on approved domains only. The model never sees it. •Self-hosted sandboxes (May 19). The agent loop stays on Anthropic’s infrastructure. Tool execution moves inside your perimeter. Cloudflare, Daytona, Modal and Vercel are supported out of the box. •Memory graduated from research preview to public beta. Only MCP tunnels and dreaming are still gated behind an access request. Pricing almost nobody quotes correctly: standard token rates plus $0.08 per session-hour. Runtime is metered to the millisecond and only accrues while the session status is “running.” Waiting on your reply or a tool confirmation costs nothing. Now the part the hype threads skip. Managed Agents is stateful by design. Conversation history, sandbox state and outputs are stored server-side. Anthropic’s own docs state that this makes it ineligible for Zero Data Retention and for HIPAA BAA coverage. So every “finally, a secure alternative to hosting agents yourself” take has it exactly backwards. You get sandboxing, scoped permissions and execution tracing. You give up data retention guarantees. For legal, health and financial workloads that trade is the whole decision. Old way: build the loop, the sandbox, the scheduler and the secret vault yourself, then ship in months. New way: define the agent, point it at a cron, ship in days, and read the compliance page before you move anything sensitive through it.

• 
  TENJI (@ineedtendies) reported 21 minutes ago

  Stayed up late into the night fixing bugs and making sure launch ran as smoothly as possible. Everything was stable when I finally got some sleep, but I woke up to reports that some users are having trouble connecting to the site. We recently added Cloudflare protection to the domain, and I’m already looking into whether that’s causing the issue. I’m aware of it and working on it now. Sorry to anyone affected. Stay tuned for updates.

• 
  lacking branding (@cutetopop) reported 22 minutes ago

  I built it for work so I don’t think I can just host it but hey maybe I’ll ask the IT guys. I think I could make it a Cloudflare worker real easy and throw on the never used art site.

• 
  Fadeout Society (@fadeoutsociety) reported 23 minutes ago

  Leak site ignoring your DMCA? Check who hosts it (whois the domain, then the IP). Hosts and CDNs have their own abuse desks and legal exposure - a notice to Cloudflare or the upstream host often works when the site itself never answers. Always work up the chain.

• 
  Xander Mol (@xahmol) reported 24 minutes ago

  @DefenceForceOrg Only thing that I do know is that I had similar issues few years ago adding Cloudflare to my then Wordpress site. No clue how I solved it in the end other than that it did cost lots of frustration and Google research.

• 
  Koushik Sahu (@laliability) reported 24 minutes ago

  Cloudflare stores data related to HTTP traffic and DNS that can be used to understand global internet trends. They have a global network of 330+ points of presence worldwide. They used this data to study the internet trend during WC 2026. Here's all that you need to know. 🧵2/n

• 
  Dr. Gonzo (@N3DrGonzo) reported 28 minutes ago

  @AnisDrawn PSN goes down more often than CloudFlare. And CloudFlare has an outage QUOTA. But yes, please tell me how an all digital ecosystem for a console is a good thing.

• 
  CEOofShitposts (@CEOofShitcoins) reported 31 minutes ago

  @satinvictus Yeah, had another one today too, it just links the rules, but doesn't tell me which one was broken. Not sure if it's actually a lock, or just some cloudflare thing.

• 
  Tiago Yamada Rodriguez (@YamadaLuke21) reported 34 minutes ago

  Content from today’s P0 Space: This was AI-extracted, so apologies for any errors. Below is a timestamped summary of the P0 Systems AMA, where Cory answered most questions, focusing on information relevant to investment decisions. 1. Revenue, Users, and Business Model [11:55–12:15] Users: P0 has over 300,000 free users and about 30,000 paid P0 Go subscribers. [27:25–27:40] Pricing: P0 Go costs $5 for the first month and $10 thereafter. [12:20–12:40] Payments: P0 supports Thai local payments, Alipay, and cards through services such as Stripe. Settlement takes several days to 30 days. [34:10–34:30] Pivot: P0 began as a token launchpad but pivoted into AI gateway and infrastructure. 2. OpenRouter Performance and Advantages [02:10–02:30] Cloudflare: P0 became Cloudflare’s largest open-source model customer and consumed 2.5 billion tokens, prompting pressure for price increases. [08:00–08:25] Traffic: P0 processed 200 million tokens through OpenRouter the previous day. Including direct usage, volume exceeded 2.4 billion tokens in 24 hours. [03:30–04:00] Supply constraints: Even Z-AI, developer of GLM models, reportedly cannot fully handle P0’s demand of 50–60 million tokens per minute, creating a supply shortage. [64:45–65:10] Product advantage: P0 Go offers limits about nine times higher than OpenRouter’s. 3. Proprietary GPUs and Compute Business [03:00–03:10] GPU procurement: P0 secured 32 B200 GPUs for an estimated $2 million. [13:50–14:15 / 14:50–15:00] Location: P0 is building a server farm in Guangzhou, China, due to strong grid infrastructure and lower electricity costs. [19:00–19:40] Asset value: P0 is deploying H200s and B200s. Each unit is said to be worth about RMB 4.6 million, or $680,000. [11:00–11:20] Margins: Owning infrastructure would eliminate OpenRouter’s 5% fee and providers’ 20–40% margins, significantly improving profitability. [27:50–28:00] Timeline: The Guangzhou GPU farm was expected to begin operating the following week. 4. Partnerships and Potential Partners [08:35–08:45] Tencent: P0 uses the HY3 model through Tencent and is negotiating a discount above 40%. [20:00–20:30] SpaceX AI: P0 claims a $1 million commitment and is waiting for SpaceX’s new Flash model. [23:00–23:25] Solana Foundation: P0 has a group chat with Solana Foundation members and remains in contact. 5. $P0 Token Economy and Value Accrual [32:45–33:15] Token as equity: P0 treats $P0 as equivalent to company equity. A VC wanting to invest would need to buy tokens on the open market. [39:10–39:20] Liquidity: P0 plans to add about $10,000 in liquidity the following week. [39:55–40:15] Value accrual: Since simple buybacks could be sold into, P0 is considering USDC staking rewards similar to dividends, though legal structuring remains unresolved. [42:40–43:00] Team incentives: Staff are paid in USD to avoid unnecessary token-selling pressure. 6. Roadmap and Long-Term Vision [29:05–29:30] Robotics: P0’s long-term vision extends beyond LLMs toward cloud-computing infrastructure for robots. [31:30–31:40] Automation: P0 targets markets where workers earning about $2–5 per hour could eventually be replaced by robots. [51:00–51:15] Development: P0 aims to become more than a reseller by operating its own AI lab and fine-tuning models internally. Speaker Attribution Cory, representing P0: The user figures, GPU procurement, Chinese infrastructure, profitability, and token-as-equity policy were Cory’s specific claims. Host: The host discussed rumors of a possible $10 billion OpenRouter acquisition at [25:10] and compared P0 with other infrastructure protocols at [22:15], prompting Cory to explain P0’s moat. Specific Commitments and Deadlines - GPU farm launch: The following week in Guangzhou. - Liquidity addition: The following week, about $10,000. - Funding: Of the roughly $2 million required, the remaining $1.7 million was reportedly covered by initial capital and funding from VCs and friends [12:55].

• 
  Michael Hartwell (@Teresa_says_) reported 35 minutes ago

  I almost ignored these charts — then the same signal appeared across all of them. $RBLX — Roblox — Don’t buy $DUOL — Duolingo — Don’t buy $APP — AppLovin — Don’t buy $HIMS — Hims & Hers — Don’t buy $SNOW — Snowflake — Buy at $252-$261 $DDOG — Datadog — Buy at $242-$250 $NET — Cloudflare — Buy at $251-$259 $SPOT — Spotify — Buy at $453-$468

• 
  aberba (@aberba) reported 38 minutes ago

  @0xRapid @dok2001 @OpenRouter The problem is Vercel provides a significantly better DX that Cloudflare and and they know it. Vercel is henceforth more expensive. Depends on what you're looking for. I use both. Wish I could use just one.

• 
  Dbug (@DefenceForceOrg) reported 39 minutes ago

  @xahmol I have no idea what the problem is Oo Could be the phpbb forum config, could be some apache parameter, could be the DNS parameters in CloudFlare or my hosting company...

• 
  Name Groove (@Name__Groove) reported 41 minutes ago

  @SpaceshipStatus Site appears down to me - cloudflare errors

• 
  mrusu.jp (@mrusu_) reported 42 minutes ago

  this cloudflare crap is messing me up technically done though

• 
  Swamp Thing (@driftinj) reported 45 minutes ago

  @NateSilver538 It really blames anyone but itself. Oh that is definitely a Cloudflare problem. Gmail is definitely doing this incorrectly. Clearly, Supabase built their auth logs in correctly.

• 
  Jake Gevorgian (@JakeGevorgian) reported 50 minutes ago

  @_slowstacking @FinancialPhys It's not like I'm not putting money in web. I spend $30k worth of time as I'm also tek savvy and do these seo stuff myself. What I've missed big time was cloudflare integration and schema. But even with that, I see companies who pop up on google first pages that I've never heard of in woodworking communities I am in... Btw, I don't think "competitor" is a valid term in this day and age for craftsmen in wood. It's not like we're selling car parts or plumbing supplies imported from China. :) this is what I'm stressing about---Google should focus on local American creators for people who look for local makers.

• 
  Sumedh Bengale (@ssbengale) reported 50 minutes ago

  @anupamrjp Cloudflare but they support a lot of TLDs I'm also scared of the day they decide to ban my account for whatever reason and eat up my whole portfolio

• 
  LilRhodySpecial (@rhody_special) reported 51 minutes ago

  @burcs Who wants a version stripped of data stealing trackers, debuggers, backdoors etc. Cuz they all up in that *****!!!! Believe that!!!! I will never use cloudflare but people who do shouldn’t have to settle for a product faking the funk!

• 
  jeli beli (@nftgamingnoob) reported 52 minutes ago

  @cooldown_sol @Pumpfun @Cloudflare damn

• 
  avyonette (@avyonette) reported 56 minutes ago

  @IntCyberDigest I already get captchas everywhere, even when not using a VPN (**** Cloudflare) At this point I'm just going to stop using the internet entirely

• 
  Graeme (@GraemeVIP) reported 57 minutes ago

  @IntCyberDigest They still allow accessible options for people who can't do it. Google Captcha V2 was terrible for disabled people and the visually impaired. V3 solved that. This is unnecessary. You can beat bots with a honeypot field and Cloudflare, you don't even need Captcha V3.

• 
  Raunak (@raunakdoesdev) reported 1 hour ago

  @samertm @Cloudflare their global control plane actually not a specific region for a specific service

• 
  Jim Liu (@jiahanjimliu) reported 1 hour ago

  $NBIS False Claims On $CRWV @pequityresearch is one of my favorite accounts on X. He is interviewed a former Nebius senior finance personal who is feeding him false information. 1. "The problem of CoreWeave is their business model is sign a contract and then allocate data centres, then allocate GPUs and buy the GPUs they don't have and then start project management." Coreweave has a 99B backlog (1). They are buying datacenter capacity and GPUs as fast as possible. What they waiting on is powered shell capacity. This made-up narrative of some slow bureaucratic process at Coreweave is bull. $CRWV currently leads both $NBIS and $IREN in absolute numbers revenue growth. 2. "The majority of their revenue, maybe 95% or more is still Hopper revenue." Coreweave does not reveal their fleet numbers but their Q4 2025 earnings call reveals they "delivered more than 50,000 Grace Blackwells" (2). Coreweave in Q1 2026 had 2.08B quarterly revenue - do the math on 50,000 Blackwells and the 95% number is obviously bull. Having kept track of Coreweave, I roughly believe only 33% Coreweave's revenue is hoppers. Coreweave is ahead of both $NBIS and $IREN on getting the latest GPUs. In fact they were the first to bring up Vera Rubin. 3. "CoreWeave’s revenue per gigawatt has decreased over recent quarters because older contracts signed during peak Hopper pricing are rolling alongside newly added capacity at lower rates." The "old contract Hoppers" are now fully paid off. These are the machines bringing in cashflow. For any Neocloud and Hyperscaler, all the new machines are cashflow negative for 3 years. The math of Coreweave's revenue per watt decreasing is so badly made up too. Coreweave has been increasing revenue every quarter of which the newer capacity is increasing revenue per watt. Maybe the old contracts rolling over decreases the rate of increase but that's the worst way to cherry-pick. 4. "Nebius anticipates that hyperscalers (like Microsoft and Meta) will only utilize neoclouds as stopgaps for five to eight years until they fully internalize their own infrastructure." Go look at Coreweave's customers - it's full of enterprises: Cloudflare, Inflection, Cursor, Augment, NovelAI, Rev, Chai, Jane Street, WaveForms AI, Decart (3, 4). 5. "However, memory accounts for a relatively small percentage of overall unit economics (e.g., roughly 2% to 3% historically), meaning even a 2x-3x price spike only minorly impacts contribution margins." Look at how much of a GPU Rack is HBM Cost. 2% - 3% historical memory cost is meaningless here when HBM is about to have a prolonged elevated cost that's. 6. For a standard 50MW greenfield powered shell, outfitting and bringing servers online takes about two months, while larger 300MW sites with roughly 90,000 GPUs take 3 to 4 months. Nebius Vineland site has taken 15 months and counting. 7. "Neoclouds actively target high capacity utilization rates exceeding 90% (often 94% to 95%) on external customer data centers to maintain competitive pricing against hyperscalers, who typically settle for 75% to 80%." No Hyperscaler has released their utilization rate but 90% utilization rate on Nebius means billing rate not actual GPU utilization rate. 75-90% GPU utilization is possible for subset of periods during training but for inference, this is impossible over a sustained period as HBM and tensor cores are used at different percentages for different tasks. Most of the time, either the HBM or Tensor cores are underutilized. For them to be perfectly synchronized is luck.

• 
  Igor bedesqui (@bedesqui) reported 1 hour ago

  @hate_santi @AmpCode @t3dotcodes The ios app is in private beta, but for now you can use the --serve command to expose your T3 code server to other devices in the same network. Then, by using something like tailscale or cloudflare tunnels, you can access the web portal from your phone regardless of network

• 
  TezaApps (@TezaApps) reported 1 hour ago

  @idunnstudio Good catch running the audit anyway. Stripe's restricted keys are the other half of that fix, a leaked key that can only read is a bad day instead of a disaster. Mine lives only in a Cloudflare Worker so the client bundle never even sees it.