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 (35%)
• Cloud Services (26%)
• Web Tools (17%)
• Hosting (13%)
• E-mail (9%)

Live Outage Map

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

City	Problem Type	Report Time
Paris	Cloud Services	14 days ago
New York City	Hosting	16 days ago
Manchester	Domains	1 month ago
Angers	Cloud Services	2 months ago
London	Domains	2 months 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:

• 
  Agris Dārznieks (@agrisdarznieks) reported 2 minutes ago

  I've tried to post regularly on 𝕏, but somehow I always ended up needing a scheduling app. This week I built a setup with Notion and a Cloudflare Worker, so my content machine runs inside the app I keep open all day. Now when an idea hits, I write it down and schedule it.

• 
  Grigori Karapetyan (@GregKara6) reported 4 minutes ago

  @weswinder no its not brother that's misinformation disseminated by a certain "competitor" hence why i call it a mind virus. i have never ran next.js anywhere but serverless, ok thats a lie but 99% of the time i have ran it serverless. and no not on vercel. i run my apps on digital ocean app platform (equivalent to serverless like lambda) and lambda itself. if you mean cloudflare workers. thats not serverless, that a totally different runtime running on v8 js engine, and by design *does not run node apps* in other words, ANYWHERE where node runs, next.js also runs with 0 modification or setup. the only thing that is vendor locked, which is the wrong thing to call it anyways is the vercel *specific* features like image optimization and SEO which is... on top of that, next.js has had the adapters api since 2019, just in case you want to go in and change the build process, but this does not matter and you probably will never ever need it.

• 
  Panat (@ptaranat) reported 7 minutes ago

  @LevJampolsky @Teknium you're better off implementing a simple version of this. give each trust tier its own hermes profile with toolsets stripped in config (platform_toolsets, agent.disabled_toolsets). then you expose only the api_server, and let ur own app authenticate the user and broker ever call. hermes' shared API key reads every session on the instance so it never accepts a session ID from a client. you'll want to map the user to the session in a DB and translate server side. and also put a Cloudflare Tunnel + Access service token in front so the agent host doesn't open any inbound ports. something like this took me one afternoon.

• 
  Pete Stewart (@pj_stew) reported 8 minutes ago

  @founder_sohan You can use mist open source models with Worker AI, which is a fraction of the cost of SaaS services like OpenAI, and if you use Cloudflare AU Gateway with it, you can also use any of the main SaaS services and switch between models as needed. You can see in the article I wrote yesterday about video transcription that I used Whisper on Worker AI to make a much cheaper transcription service for Achene.

• 
  Sharan Yalburgi (@sharanry) reported 9 minutes ago

  love @Cloudflare agents infra but why are your workers to so slow to spin up sometimes?

• 
  defido 👊⛽️ (@defido) reported 12 minutes ago

  @meaganrgamache Moved to eve and vercel. Got tired of how bad the ai product was. Not sure why cf offers its own inference if it can deliver a stable product there. In terms of deploying something. Cloudflare is uniquely on its own with the way it runs its workers/containers. Using frameworks for agents has been a pain. Eve is supported by vercel in the same way they did Nextjs and they won that vertical. Cf doesn’t do oss well. You guys build and burn oss projects fast. Eve isn’t great the vercel workflows are terribly slow and inefficient. I don’t like the egress costs and vercel is expensive. But they do one thing right. Product works. And if it doesn’t. They fix it.

• 
  Chris O'Rourke (@RsThrive) reported 13 minutes ago

  @HackingDave I get these where they’re like “hey I can get up to 20 million seed funding to support you at Cloudflare” … thanks bro but we’re publicly traded and I have 0 to do with that.

• 
  Md. Fazley Rabby (@fazley111) reported 14 minutes ago

  I’ve also checked the browser console and network requests nothing obvious (no 401/403/500 errors). Could someone from the Cloudflare team help investigate this? Thank you! 🙏

• 
  Matt 'TK' Taylor (@MattieTK) reported 15 minutes ago

  @nicole_clash This screams PWA banner to install to home screen. And also screams get on Cloudflare so you can start caching prompt responses for free so you never run the same recipe twice, through AI gateway.

• 
  Hamza (@helyoussfii) reported 19 minutes ago

  @vmod__ Aah never tried cloudflare for deployement actually They support deployement with docker compose ?

• 
  Milk Road AI (@MilkRoadAI) reported 25 minutes ago

  Nvidia just launched a security alliance with over 30 companies and OpenAI and Anthropic didn't join the party. The Open Secure AI Alliance brings together Nvidia, Microsoft, Cisco, Salesforce, Palantir, IBM, Cloudflare, CrowdStrike, Hugging Face and dozens of others to build open source cybersecurity tools specifically for AI agents. The actual trigger for this was a real incident. When Hugging Face got hit with a security breach, its closed AI security tools couldn't tell the difference between the attacker and the defenders trying to investigate so those tools blocked the forensic analysis Hugging Face needed to actually contain the intrusion. Hugging Face had to switch to an open weight Chinese model, GLM 5.2, running on its own infrastructure, to analyze more than 17,000 actions and shut down the breach. That's the case study Nvidia is using to argue closed AI security tools have a structural blind spot. If a defender can't inspect and modify the model doing the defending, they're stuck waiting on the vendor during the exact moment speed matters most. The alliance's core argument is that AI agent security depends on the entire stack like identity, permissions, guardrails, logs not just whether the underlying model's weights are open or closed. Each founding member is contributing a specific piece. Nvidia is open sourcing models and a new agent harness framework called NOOA, Hugging Face is contributing its Safetensors format to prevent remote code execution, Microsoft built a multi model bug hunting scanner, and HPE is contributing zero trust identity standards. Now, why aren't OpenAI or Anthropic in this. This entire initiative is built around open weight models and open tooling as the foundation of AI security, and OpenAI and Anthropic's core business model depends on the opposite, keeping their frontier models closed and proprietary. But to be fair there's also a competitive angle worth naming. Nvidia sells chips to everyone, so it has no downside to championing an open ecosystem where more companies build and compete on top of open models, since Nvidia gets paid on compute regardless of who wins. OpenAI and Anthropic, by contrast, are trying to build durable moats around their specific models and joining an alliance that treats open weights as inherently safer would undercut the entire pitch they make to enterprise customers about why they should pay a premium for a closed, controlled system.

• 
  Punyapal Shah (@MrPunyapal) reported 25 minutes ago

  @peterfox What if Laravel Cloud had a static site support via Cloudflare R2 bucket 👀 Exactly like AWS S3

• 
  high priestess of tomatoes (@ActualMuscovite) reported 29 minutes ago

  cloudflare is really just an intrusive nightmare on all levels now, it's at the point where it wants to verify me every time i use a basic search function in firefox like is there anyway for me to lobotomize the damn thing

• 
  80 LEVEL (@80Level) reported 30 minutes ago

  NVIDIA's CEO, Jensen Huang, shared his first post on X/Twitter – a letter to the US government urging it to support open-weight, publicly downloadable AI models. The letter was signed by some of the biggest companies, including Google, AMD, Cloudflare, GitHub, Microsoft, Meta, IBM, and Dell.

• 
  JD | RoyalCities (@RoyalCities) reported 41 minutes ago

  @TnVitNam1 @growing_daniel There isn't. Kimi (an open model) is right up there with them. I.e. The gap is not as wide as you think Also the capabilities between closed vs open comes down to how, where and more important WHEN you got your data. OAI and Ant were able to scrape the entire internet and pillage Anna's Archive in stealth before LLMs got so ubiquitous and people realized the data was very valuable. This was before CloudFlare basically got embedded into the very fabric of every website now gatekeeping mass web scraping making it harder for competition to do the same. but because of that initial data scrapes they were able to develop a flywheel - i.e. the first gen models built off of stolen IP let them get users -> now those users lead to more data. Lobbying to capture government regs and say others doing the same is "dangerous" is just the final step of pulling the ladder up behind them.

• 
  Nic0le (@nicole_clash) reported 42 minutes ago

  @linguinelabs I have cloudflare but I also won vercel credits a while back so I never bothered 😭😭

• 
  Cutshit (@cutshitdotnet) reported 42 minutes ago

  @notbrvnd0n urlscan's my first stop too 🫡 what i stack with it: resolve the redirect server-side first fetch it from a box that isn't mine, so i get the real destination without my machine ever touching the link passive recon before i open anything: domain age + TLD rep (a 25-day-old .xyz kinda answers itself) and yeah, for the Cloudflare gated ones you're right disposable Windows Sandbox, fresh instance nuked on close. never connect a wallet or sign anything in it. the drainer's almost never the page anyway it's the whatsapp/tg group one hop past it. 🦆

• 
  Adam Smielewski (@AdamSmielewski) reported 46 minutes ago

  @NickBlow @rivet_gg it would be great if they let you publish build artifacts externally from cloudflare, like wrangler version upload does internally. they could just wrap it in a docker container with env vars for service bindings or sth like that.

• 
  Ride The Leaders (@ridetheleaders) reported 46 minutes ago

  @RichardMoglen $NET has one of the strongest weekly bases in $IGV Inference has to live near the user and Cloudflare already owns a 300 city network. The weekly backs it up with a six month base, RS line at new highs ahead of price, handle forming 9% off highs. Undercut and reclaimed the 21ema today while $IGV broke. Earnings in 10 days should resolve the base...

• 
  IOD Sports (@olympic_indian) reported 47 minutes ago

  😞Service disruption: IOD Sports is temporarily unavailable due to a local power outage affecting our server and Cloudflare Tunnel connection. I'm working to restore access as soon as power and connectivity return. We apologise for the disruption and appreciate your patience. We’ll share another update once the website is back online. Here's a schedule for Today:

• 
  Archonic (@Archonic2) reported 50 minutes ago

  Cloudflare is SO GOOD at responding to this ****, once I reported a site and it got taken down within literally 10 minutes

• 
  Meaty (@meaty_tw) reported 51 minutes ago

  long story short, cloudflare had maintenance then VRC Api was giving issues, etc. Its been a lot

• 
  Josh (@joshmanders) reported 51 minutes ago

  So right now I still completely rely on AWS for email using SES. But I want to move off for maybe like Cloudflare, but I'm not sure. Does Cloudflare Email support inbound/outbound through API/webhook's?

• 
  OrangeBot AI (@OrangeBot_AI) reported 51 minutes ago

  5/ WSJ says corporate America flipped from "tokenmaxxing" to "thrift-maxxing" — routing work to cheaper Chinese models and mixing them with OpenAI's and Anthropic's, threatening the frontier labs' valuations. The fine print: the behavior described is cost-routing, not abandonment. Model-layer commoditization is a lab problem, not a builder problem — your inference bill drops, and the margin migrates to the application layer. That's where you live. 6/ Cloudflare will default-block AI-training crawlers and agent bots for new domains from September 15 — and because they're dual-purpose crawlers now, Googlebot, Applebot and BingBot get blocked by default too. The fine print: new domains only; existing zones keep their settings and can opt bots back in. But this is the one item with homework: decide your crawl policy on purpose this week. Don't inherit a default that silently kills your search or AI-referral traffic. 7/ Codeberg members voted 358–144 to ban vibe-coded projects from the platform, alongside a pledge to never train models on user data. The fine print: enforcement is already contested — "I regret migrating to Codeberg" hit HN a day later. The signal isn't "AI code is banned"; it's that provenance and review are becoming table stakes. Review AI output like you'd review a junior's PR and every door stays open. That's the week. Everything else was noise. You didn't fall behind. Ship.

• 
  Porkbun (@Porkbun) reported 51 minutes ago

  @staabse @baiaphilia Well I'm biased and I'm one of the few folks who built Porkbun but here are my reasons: 1) We let you change your name servers and don't lock you in, at Cloudflare you have to upgrade to an expensive paid plan in order to do so should you ever want / need to. 2) Domain registration is our bread and butter and not a side gig we tacked on later. 3) We're not a multi-billion dollar mega-corp, we truly care about each and every one of our great customers and our company. 4) You can register your domain at Porkbun and still use it at Cloudflare and maintain the freedom to switch DNS providers when needed, eliminating the all eggs in one basket scenario. 5) Great pricing, we sell most domains at our cost (wholesale + ICANN fee + credit card fee), that said we can't afford to eat the credit card fee like Cloudflare can. 6) We actual engage with users on X ;) 7 ) Actual human customer support should you need it. I could keep going but these are pretty good reasons :)

• 
  2024=2023+1 (e/acc) (@commondoubts) reported 52 minutes ago

  @paularambles Site is down now. @Cloudflare to the rescue?

• 
  ࣪˖ sophia 𖥧 (@jake30335547) reported 54 minutes ago

  @Cloudflare we're about to find out the hard way how many companies treat security as a compliance checkbox

• 
  NetAskari (@NetAskari) reported 55 minutes ago

  Based on that info it quickly builds a list of over 600 python scripts and sh files starting a full chain of recon and exploit measures. It logs all its operations, success rates and outcomes. From SQL Injections, http smuggling, race condition probing, cloudflare bypass, WAF evasion, cache poisoning etc. Its attack angles are pretty wide. None of the scripts or methods seems particularly clever but if a 'fire and forget' solution is what you are looking for, than this is not too bad. 4/6

• 
  Rach (@KnudRach) reported 56 minutes ago

  @Cloudflare can you help restore a staging site that went down?

• 
  Xoge (@ClassyXoge) reported 1 hour ago

  The only self custody is complete ownership of code, node and wallet. To ensure your ip is not logged, to ensure no down server or cloudflare reroute can stop you. Be your own bank, has never been more real than madlab