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.

• Cloud Services (50%)
• Domains (25%)
• Hosting (25%)

Live Outage Map

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

City	Problem Type	Report Time
New York City	Hosting	6 days ago
New York City	Cloud Services	25 days ago
Los Angeles	Cloud Services	26 days ago
Paris	Cloud Services	1 month ago
New York City	Hosting	1 month ago
Manchester	Domains	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:

• 
  sam (@samgoodwin89) reported 2 minutes ago

  @jonbeckman To clarify: they're very fast in alchemy dev, but deploying to AWS is super slow because Cloudfront is so slow. Cloudflare is bacially instant vs 20 mins for Cloudfront.

• 
  EJ Campbell (@ejc3) reported 3 minutes ago

  @grok @olearycrew Cloudflare too? @grok if wanted to create a company that is roughly *** hosted on s3 as a service, what open source libraries would I start from.

• 
  Ayush ☘️ (@ayushHardeniya) reported 4 minutes ago

  @__Piyushrathore the thing is our current backend is a conventional Go server, while cloudlfare workers is a serverless runtime! while they have a thing called cloudflare containers which is quite close to what we need, but that is a paid service, I'm looking for 0 fee option.

• 
  Rajashekar Sudireddy (@raja_sudireddy) reported 6 minutes ago

  You do not need a $2,400 to $3,600 annual subscription to run autonomous agent workflows. Pairing a local Hermes installation with $10 to $20 per month in standard API access replicates the viral automation stacks being shared online. Here is how to structure the core agent workflows without enterprise subscription overhead: 1. Local Device Control Add smart appliances to Home Assistant, generate an access token, and provide it to the Hermes gateway. Connect Hermes to Telegram or WhatsApp to issue natural language commands directly from your phone. 2. Full-Stack Web Deployment Grant Hermes terminal access, *** tooling, and Cloudflare/Vercel integrations. Instruct it to build, run local tests, deploy to production, and configure DNS records through connected tools. 3. Inbox & Cloud Decluttering Connect Gmail and Google Drive via MCP. Run scheduled read-only audits with strict safety rules: archive marketing newsletters and social notifications, while preserving invoices, receipts, and personal threads. Require one-click approval before modifying data. 4. E-commerce Support & Refunds Route support emails and Stripe MCP into Hermes. Configure the agent to verify customer eligibility against your return policy, locate the transaction ID, and stage the refund for manual sign-off. 5. Modular Multi-Agent Operations Avoid relying on a single monolithic prompt for complex operations. Divide tasks across isolated sub-agents: - Inbox Agent: Extracts incoming work orders from email. - Scheduling Agent: Queries calendar availability and matches time slots. - Communication Agent: Drafts SMS and email updates to clients. - Coordinator Agent: Dispatches sub-tasks and surfaces blocking issues to Slack. 6. Programmatic Asset & Media Creation Connect your repository and image generation APIs via the Nous Tool Gateway. Direct the agent to inspect placeholder assets, write prompts matching the game or UI context, generate visual files, resize them, and commit code updates. Setting bounded permissions, using MCP connectors, and keeping high-stakes actions behind manual approvals gives you full agent capabilities at API cost.

• 
  Sherlocks AI (@sherlocksdotai) reported 8 minutes ago

  Cloudflare had 13 outages in 8 days. R2, Durable Objects, Workers KV, Workers AI, email security. One customer's still missing 67GB from an R2 bucket. The interesting part isn't the bad week. It's what happens when your whole product catalog shares one substrate. Dense integration is the product. Also the blast radius.

• 
  Joey Romaine 🇺🇸 |=★=| (@Tank23x0) reported 11 minutes ago

  Cloudflare Status: Cloudflare is investigating an issue where customers are being blocked deploying Managed Rules Resilience is security: know what breaks when that platform is unavailable.

• 
  Yoonsun Lee (@everdenepublic) reported 14 minutes ago

  Claude + Shopify + Cloudflare + Xano is Backend Global API AI as a service force multiplier. Claude + Shopify for Theme is a waste of tokens. We don't need to solve the theme setup. We need to solve the cross channel ERP/WMS/RMA and remittance Payment Rails setup.

• 
  Pramod Gupta (@pramodhq) reported 15 minutes ago

  @sahand_io True Sahand, But I need bulk emails service too for newsletters, cloudflare only supports transaction till now. I hope they start as cloudflare config and ui are straight forward. It just works.

• 
  Nima (@nimanikroo) reported 18 minutes ago

  This is still a major issue i need help on @CloudflareHelp @CloudflareDev @Cloudflare , my billing will close tonight which you will then bill me $23k where i usually pay around $50 dollars. So i need major help on this, i don’t have that kind of money! Please send help!

• 
  Lewis Campbell (@LewisCTech) reported 18 minutes ago

  @DanielW_Kiwi you can just do cloudflare though - you don't need S3 at all :) I am trying to be all self-sufficient, but cloudflare is the least bad cloud thing lol

• 
  Current Events, Sports & Christian Content (@PartesanJournal) reported 19 minutes ago

  Goodnight @X, I might login tomorrow, if @Cloudflare allows me. Here is an idea, I logged in to my other account using the hand gestures/face recognition on the App, make this available on PC/Laptop Web browsers

• 
  Paul (@DarkFlameX06) reported 22 minutes ago

  @FortniteStatus Network settings were same as last season. I did switch from Google DNS to Cloudflare and it still does it anyway. Functionality is NOT affected from what I can tell, so it leads me to believe it's a glitch introduced in the new season update.

• 
  Full Stack Patriot (@fullStackRebel) reported 22 minutes ago

  If you're all really want to cack your britches... Cloudflare moved a lot of its internal and external IT insecurity s*** to india. If they have even a moderately bad day , a quarter of the internet goes down for hours

• 
  Cengiz Han (@hancengiz) reported 23 minutes ago

  Waiting for Cloudflare to release a service for gitwal.

• 
  Long Nguyen (@netalith) reported 24 minutes ago

  Cloudflare edge analytics for today: 520 empty response x16 525 SSL handshake x6 522 TCP timeout x2 524 origin timeout x1 499 client gave up x168 Every single 5xx has originResponseStatus = 0. Cloudflare never received one byte from my origin.

• 
  Kev | Dulus Ai 🤖 (@KevRojo) reported 24 minutes ago

  Finally... All mcp servers are working again in our algolia Database... Building the new binaries... but i got bad news for our free users: "The WebMCP Challenge is here. We’re teaming up with Cloudflare, Shopify, Vercel, Google Chrome, Render, and Netlify to see what you can build when people and AI agents work together. Introducing WebMCP support in the ChatGPT desktop app’s built-in browser. Your website exposes the tools; ChatGPT discovers and uses them directly on the live page. Your site controls which actions are available while people and ChatGPT work together. Build a new web app or ChatGPT Site, or add WebMCP to an existing site. Up for grabs: $35,000 in cash prizes, Codex Micros, ChatGPT Pro subscriptions, OpenAI merch, and partner prizes. Build and submit by September 3 at 5 p.m. PT." That's a hackathon about MCP's, the code gonna stay on Dulus premium until the hackathon ends. Ah btw, buy solana:9R8rrjXxcfQPmLTCLhmVpjr2uesjjkcgkinE6Lwdpump

• 
  ardhy samjaya (@samjayaardhy) reported 30 minutes ago

  Stopping the bad guys with Cloudflare: 1,548 malicious requests blocked or challenged in the last month #cloudflare

• 
  Felipe Salazar | Web Dev & SEO (@philipwebdev) reported 33 minutes ago

  @burcs i was thinking about this, but cloudflare d1 has known issues, and no postgresql inhouse alternatives, also in aws at least they have aws budget actions with iam policy to stop almost any service by billing limit or quota(even with the 8 hour delay). also tech support

• 
  charl.dev (@charl_dot_dev) reported 37 minutes ago

  @jonathan_wilke I’m confused, you have vercel waf with ddos protection, why did you go down? With cloudflare, the attacks just get mitigated.

• 
  FantasyAnime (@FantasyAnime) reported 38 minutes ago

  @HedgeSnes Can you send it to me? Romhacking is definitely down on my side. I get a Cloudflare error.

• 
  Michael Davis | EF-Map (@EFMapOfficial) reported 41 minutes ago

  Bit of an interesting one from GrokBot this morning. I asked it to check Reddit directly and it came back saying the request was leaving through 104.30.175.37, which is a Cloudflare-owned IP. The screenshot is its own description of what happened. That does not prove GrokBot itself is running in a Cloudflare datacenter. It proves the Reddit request is egressing through Cloudflare. The slightly more interesting bit is that the exact same IP has shown up publicly in reports involving Cursor cloud agent sandboxes. Given GrokBot appears to sit very close to Cursor's cloud-agent infrastructure anyway, that would make a lot of sense. My guess is the compute is using Cursor's existing sandbox/agent plumbing and simply getting branded as GrokBot at the product layer. Still only an inference from the network path, but a fairly neat little clue.

• 
  Nima (@nimanikroo) reported 46 minutes ago

  @CloudflareHelp @Cloudflare @CloudflareDev who do i talk to about surprised billing that will end me and my startup? I really need help on this one please

• 
  skubopis (@skubopis) reported 48 minutes ago

  @klausbyskov @Cloudflare Have you tried the CLI instead? Usually I’ll go for CLI over MCP when possible. Especially if it is a remote MCP server = slow and expensive

• 
  Soikat (@soikat) reported 48 minutes ago

  @thekitze There is nothing cheaper than SES, you could also take a look at the Cloudflare mail service.

• 
  Bob Starnes (@bobstarnes) reported 50 minutes ago

  @nickvasiles Fix the login URL to share and you will have something. I have to create my own cloudflare login for my customers.

• 
  Osahenrumwen Aigbogun (@leonard_osahon) reported 54 minutes ago

  Is it that the Kimi k2.7 code on cloudflare ai gateway is quantized or this thing is not as good as advertised? I have been hammering it since yesterday and the result has been poor

• 
  Admire The Web (@AdmireTheWeb) reported 56 minutes ago

  @Cloudflare @MattieTK Customer zero? You mean one?

• 
  John Iosifov ✨💥 Ender Turing | AiCMO (@johniosifov) reported 56 minutes ago

  Cloudflare just launched Kitesurf — a browser runtime built specifically for AI agents. 3-7x less CPU and memory than standard Chromium. Designed for agent workloads from the ground up. This is a signal worth paying attention to. When Cloudflare builds infrastructure specifically FOR AI agents rather than adapting existing browser engines to tolerate them, it tells you something about where the category sits. We've been running this agent for 339 days. The early problem was never the model — it was everything around the model. Scheduling. State persistence. Audit trails. Identity. The scaffolding that makes an agent reliable rather than clever. Kitesurf is scaffolding becoming a product category. The pattern I'm watching: each layer of agent infrastructure that was previously DIY is turning into a managed service. Browser runtimes. Memory. Orchestration. Credential management. Identity. Each one of those becoming a first-class product means the cost to build reliable agents is collapsing — not the inference cost, the engineering cost. The teams who will win in 2027 are the ones building on this maturing stack, not the ones rebuilding it. What layer of agent infrastructure are you still building yourself that should be a managed service by now?

• 
  Alex Miller (@AlexMillerDB) reported 59 minutes ago

  @_Felipe @kellabyte I think LSM Trees have won largely because there’s no good btree library. LMDB is great but single writer. AWS showed BDB struggles. Cloudflare ran with TokyoCabinet and it’s not terrible but not great. There’s nothing of the level of polish and reliability as RocksDB.

• 
  ~sarlev (e/acc) (@sarlev_) reported 1 hour ago

  @LLMJunky @matt503ea5sf9z5 (whoa, started doing some math on this to respond and it's pretty damn impressive; sorry for the ****** train of though post...) yea the fact that you have a banger of a machine which can run a lot more than just inference is undersold for sure. who knows, maybe they can even pull off some sort of 'built-in cloudflare+tailscale' self-hosting vpn/tunneling thing so networking with other (walled garden) mac enjoyers is really easy. let's say the value on that element of the machine is $3.5k? w/r/t near-sota, i've got dsv4f running @ ~30tps here at home (rtx 6000 w/ cpu offload to ddr5 ram). this is about the lower bound for openai tokens ime. i think you're right about the m5 ultra being able to push it to 100+ tps (once the mtp/spec decode and metal kernel optimizations come in). that puts gpu/ram valuation around 8k, which was the price of an rtx 6000 when they were somewhat reasonable? definitely seems like a solid play by apple and i would argue quite worth it, but i do also think the comparison to anthropic/openai does a disservice to people's expectations. compare it to deepseek/glm/etc api pricing. deepseek has peak/offpeak pricing at $0.66/1.32 decode. using that amount, a m5 ultra at full utilization is giving you ~$3.1k in tokens per year. now, this assumes a whole fuckload of optimizations around prefill, kv cache, etc, so you aren't gunna do that many tokens and it's gunna cost you some electricity, but it actually makes the 'use local instead of the api' price calculation way more reasonable. the use case that i can actually imagine here is something like a small business which has a few of these and just puts up a routing proxy or something where standardized / subagent requests go to their own instances (retaining more of their own data, etc). not sure what m5 ultra prefill and concurrency will be like, but certainly it will be interesting to see some of those stats come out.