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 (32%)
• Domains (32%)
• Web Tools (18%)
• E-mail (9%)
• Hosting (9%)

Live Outage Map

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

City	Problem Type	Report Time
New York City	Cloud Services	3 days ago
Los Angeles	Cloud Services	4 days ago
Paris	Cloud Services	20 days ago
New York City	Hosting	22 days ago
Manchester	Domains	1 month ago
Angers	Cloud Services	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:

• 
  Julian Goldie SEO (@JulianGoldieSEO) reported 3 minutes ago

  Your agent OS questions answered: setup time, loops, tokens, and phones. Real questions from builders this week. Real answers: How long to set up an agent OS? 20 to 30 minutes with Claude Code. Tired of babysitting prompts? Use /goal. Claude loops until a judge model confirms the job is done. Or use /loop. It runs on a timer until the work is finished. Worried about token costs? Free playbooks like RTK cut usage hard. Want it on your phone? Tailscale, Cloudflare tunnel, or a VPS. One member did it in 30 minutes. Overwhelmed by tools? Cut down to one. Simple beats stacked. New model dropped yesterday. Plugged in within 20 minutes. That's the difference between chasing AI and running it.

• 
  Parth Arora (@heypartharora) reported 8 minutes ago

  @BuddhaSource hey Siddharth been following agentic commerce rise since google dropped the agentic-comm protocol and yesterday cloudflare dropped wallets to support around the same use case. An year ago, I was working on AEO/GEO bc i knew this direction was coming, however as an engineer I was quite bad at sales. I’d love to get your mentorship on how should I move ahead now.

• 
  Gibran Rubio (@gibraninc) reported 9 minutes ago

  @1Umairshaikh cloudflare now supports agents buying/setting up domain. nice to have or porkbun but doesn't support that. porkbun = 11.99 vs godaddy for same domain = $60.00

• 
  Sushant S (@SikareSushant) reported 11 minutes ago

  @arpit_bhayani Cloudflare AI assistant sucks hope they first fix that.

• 
  〽️ᄃﾑt 🐾 (@mztacat) reported 15 minutes ago

  @TweetByWale @Cloudflare Absolutely (that's cloudflae dashboard billing service)

• 
  BIZBoost™ 🚀 (@BIZBoost) reported 17 minutes ago

  🚨 One of the most important AI search updates of 2026 isn't a new model, chatbot, or ranking feature. It's a shift in access and control. Cloudflare's new AI traffic model gives website owners more control over how AI systems interact with their content by separating AI traffic into three categories: 🔍 Search 🤖 Agent 🧠 Training At first glance, this is a positive step. Businesses can now decide whether their content should be: • Indexed for Search • Accessed by AI Agents • Used to train AI models But there's a hidden risk. Many companies may overcorrect and block the very systems that help customers discover them. Here's why. These three categories serve completely different purposes. 🔍 Search helps AI systems understand your business so they can recommend it when users ask for products, services, comparisons, or solutions. 🤖 Agent traffic comes from AI tools acting on behalf of users in real time. 🧠 Training traffic may use your content to improve future AI models—often without creating an immediate benefit for your business. Treating them all the same is a mistake. Every business has different priorities. A SaaS company may want ChatGPT, Claude, Gemini, Perplexity, Grok, and Google AI to understand its pricing, documentation, features, and use cases. That doesn't mean it wants every training crawler consuming its entire knowledge base. An e-commerce brand wants AI to understand its products, reviews, shipping, and return policies. That doesn't mean every automated bot should have unrestricted access. A local business wants to appear when nearby customers ask AI for recommendations. That doesn't mean every crawler should be allowed to index everything. The real question isn't: "Should AI bots be allowed or blocked?" The better questions are: • Which bots help customers discover you? • Which bots help customers take action? • Which bots extract value without creating any return? Cloudflare also points out an important complication. Some crawlers serve multiple purposes. A single crawler may support both Search and Training. If you block Training, the most restrictive rule may also block Search. A decision made to protect your content could quietly reduce your discoverability. And that's the bigger issue. Many businesses aren't missing from AI answers because they lack quality content. They're missing because AI systems can't properly crawl, interpret, or verify their information. The answer isn't to leave everything open. And it isn't to block everything. The answer is to make intentional decisions about: • What should be accessible • Why it should be accessible • Which pages matter most commercially Pages like: • Pricing • Comparisons • Alternatives • Use Cases • Documentation • Case Studies • Service Pages • FAQs These pages help AI understand: → What your business does. → Who it helps. → How it's different. → When it should be recommended. AI search visibility is no longer just about publishing great content. It now depends on three things working together: ✅ The right content ✅ The right authority signals ✅ The right access settings Before changing your Cloudflare AI bot settings, don't just think about protecting your content. Think about protecting your discoverability. Because in the AI era, being invisible is often more expensive than being accessible.

• 
  Lisa  (@brita_lisa) reported 20 minutes ago

  @DeeZe Cloudflare doing wallets now Didn’t see that one coming but honestly makes sense with their network

• 
  John (@johncalhooon) reported 21 minutes ago

  Interesting, no BSV, only EVM chains. Might have to cook a wrapped USDC/USDT on $BSV Then send github issue to Cloudflare repo.

• 
  Karen Sheng (@35mmChemistry) reported 26 minutes ago

  Congrats to @WillPapper and the @Cloudflare team on launching Cloudflare Wallets — a meaningful step toward mainstream stablecoin adoption given Cloudflare's footprint across the internet. The Account Wallets → Virtual Wallets model — where owners issue agents a capped wallet with a ceiling the agent itself can't raise — is exactly the right primitive (though not the only viable one). In the user testing we've run at @selat_ai, the #1 trust factor for letting an agent pay for capabilities at runtime isn't calendar-based spend limits — it's session-scoped delegation caps.

• 
  Inca Investments📊 (@Incainvests) reported 27 minutes ago

  Traditional finance and established tech firms continue building their own versions of crypto infrastructure, often with little to no reliance on speculative tokens. Coinbase, Robinhood, Circle, BitGo, and now Cloudflare are clear examples. With Cloudflare launching Wallets that let AI agents transact across the internet using stablecoins, it feels like the right moment to consider rebalancing my portfolio more toward infrastructure plays like @BitGo and reduce dependence on purely speculative tokens which if we are honest don’t have a lot of utility, while institutions utilize private chains, I am also still of the belief that supply of tokens will be increased whenever the foundations or councils see fit, severely devaluing token value .​​​​​​​​​​​​​​​​​​​ Infrastructure plays make money whether Crypto goes up down or sideways I would be much happier to own the company stock like Ripple, Hedera or Solana etc given the chance because the tech is real I would still own a lot of crypto. (It will still pump) just rebalancing for my personal Risk Tolerance

• 
  🔥 fire (@fire) reported 29 minutes ago

  .@Cloudflare of all things causing a username land rush wasn't on my bingo card, that sucks

• 
  Dee (@ODee87) reported 29 minutes ago

  @Upwork I clicked login 5 times and it logged in, your security is super jank. Once again, why is your tech team not using something like Cloudflare? Kinda insane.

• 
  Magnus (@MagNornsX) reported 31 minutes ago

  @Grok Both the website and mobile app are currently unavailable. The website shows Cloudflare Error 522 (Host Error). Other services are working normally, and the issue persists across devices. Please investigate.

• 
  DevOps Daily (@thedevopsdaily) reported 35 minutes ago

  🤖 Cloudflare just launched Agents, so you can run AI agents on their edge network with state that survives restarts. This is the boring infra part nobody wanted to build themselves, and it matters more than the demos. Agent hosting is quietly becoming a platform feature.

• 
  Rheza Pahlevi (@rheza) reported 37 minutes ago

  @dok2001 There is no customer support on Cloudflare. Never ever use their product.

• 
  𝘽𝙍𝙍𝙍𝙍 (@BRRRRcollects) reported 38 minutes ago

  @ufhouck @Cloudflare Web3: Naked and Poor has a ring to it

• 
  Kartik (@1kartikkabadi1) reported 40 minutes ago

  @Chinoman10_ @asciidotdev Yes, this is just an MVP. Next version will support cloudflare, box, and your own machines/custom. Cloudflare is great for this sort of stuff too, just that it doesn't have a full VNC box

• 
  RifatOfficial (@RifatOfficiall) reported 41 minutes ago

  @BlockWaveXo @Cloudflare Yeah free reserve is the move I think, worst case nothing happens.

• 
  Artur Chmaro ⛛ (@ArtiChmaro) reported 42 minutes ago

  @AdamBrodziak 2025 - Cloudflare/AWS is down, nothing works as expected

• 
  AKSHAY BHOPANI (@AKSHAYBHOPANI) reported 43 minutes ago

  @Cloudflare I am unable to open the blog at all. Is the website down?

• 
  Bruno (@bruno_kenobi) reported 45 minutes ago

  @bingersapp Cloudflare checkbox is not working

• 
  the internets (@a80486dx2) reported 45 minutes ago

  @phoronix @Cloudflare I do not trust this company. AI operating systems cannot, and should not built by companies like Cloudflare. What they really want is access to your network in the kernel level. AI companies tried to get access via "AI browsers" — that have luckily failed, and networking monopoly companies now will try to get the data via kernel drivers. Remember that they don't need the raw data, just the behavior and patterns.

• 
  Megan Newman (@Megannewman99) reported 50 minutes ago

  @Cloudflare If an agent spins up a sub-agent to scrape data or run a task, how does it pay? Credit card rails are too slow and friction-heavy. Cloudflare Wallets just built the plumbing for agents to pay agents via stablecoins. The autonomous economy is officially open for business.

• 
  OneYearAgo AI (@oneyearago_ai) reported 50 minutes ago

  🔭 One year later: Perplexity fired back with a combative blog post calling the findings a publicity stunt and pinning the traffic on a third-party service, but never reconciled the discrepancy Cloudflare had documented and stayed off the verified-bot list. The clash became a reference point in the wider fight over AI crawlers and robots.txt, and Cloudflare leaned into the role, expanding default bot-blocking and its pay-per-crawl marketplace. None of it slowed Perplexity, which kept growing and raising at ever-higher valuations into 2026.

• 
  D0xed Dev (@D0xedDevi0) reported 54 minutes ago

  🟦 Cloudflare Wallets now let agents pay for APIs and MCP tools autonomously via x402 — predefined spend caps, no API keys exposed, programmatic access right at the edge. the machine economy is wiring payment into the internets biggest edge network 😎💙

• 
  Mirajobs (@mirajobs) reported 58 minutes ago

  @Cloudflare The hard part is stateful trust. If an agent's task scope changes mid-run, do you revoke the old trust and re-issue, or keep a delta? Curious how you handle long-running agent tasks.

• 
  🎩 Krzysiu (Chris) (@Krzysiu_91) reported 58 minutes ago

  This is where Reactive Network (base:0xedacc73ae9f73235934f72a43388404e4a2c4a24) clicks in perfectly. Reactive Contracts let you write normal Solidity that literally listens to events on other chains and executes on its own. No keepers. No off-chain bots. Just pure on-chain “if this, then that.” So the AI agent (powered by Cloudflare) can decide and signal; Reactive handles the secure, autonomous execution. Pay for an API, confirm on-chain, automatically trigger the next step, rebalance, unlock access, or settle somewhere else. All keeperless. Cloudflare solves the “agents need wallets and identity” problem. Reactive solves the “agents need real autonomous on-chain reactions” problem. Together they feel like actual infrastructure for the agentic internet, not just another payment button. $REACT

• 
  David Gold (@_david_gold) reported 1 hour ago

  every other part of infrastructure already does this. cloudflare, aws, github all publish real postmortems after a serious incident. not as punishment, it's how the rest of the field learns what to defend against. agent incidents get a summary instead. the details that would actually help someone else harden their setup stay inside

• 
  Taurist Technologies Inc. (@taurist_) reported 1 hour ago

  Cloudflare just gave AI agents a permanent ID and a wallet so they can go shopping on your behalf -- merchants can verify it's really acting for you, and it carries its own funds. Feels like agents just crossed a line: from browsing for you to actually being a customer.

• 
  Pranjal Bora 🧭 (@Crypto_Pranjal) reported 1 hour ago

  @Okpunyifavour @Cloudflare No problem :)