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 (47%)
• Domains (20%)
• Web Tools (13%)
• Hosting (13%)
• E-mail (7%)

Live Outage Map

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

City	Problem Type	Report Time
New York City	Cloud Services	15 days ago
Los Angeles	Cloud Services	17 days ago
Paris	Cloud Services	1 month ago
New York City	Hosting	1 month ago
Manchester	Domains	2 months 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:

• 
  Choblin (@choblin29) reported 5 minutes ago

  Holy. Anthropic is reportedly projecting $190B-$200B in revenue by 2028. That is more than 4x its $47B revenue run rate in May. And Wall Street is already looking that far ahead to help value the IPO. Bankers and investors are using forward revenue multiples, with Palantir, Cloudflare and SpaceX among the reference points. Anthropic went from a ~$9B run rate at the end of 2025 to >$47B by May. It is also projecting at least $10.9B of Q2 revenue and is on track for its first quarterly operating profit at $559M. $200 billion revenue in 2028. Absolutely freaking insane growth if they can actually pull this off.

• 
  Laurent Perrier (@maybelaurent) reported 5 minutes ago

  @Cloudflare is anyone working on this? your status page still doesn't show it's down

• 
  Milan Jovanović (@mjovanovictech) reported 5 minutes ago

  How do you handle errors coming from external APIs? Here's an example using the Keycloak REST API. If you're unfamiliar with it, Keycloak is an OAuth-compliant identity server. There's an API endpoint you can call to register a new user with Keycloak. If the provided email isn't unique, you'll get an error (409). The correct response code for this type of error is 409 Conflict. You are trying to create a "resource" that already exists. What would you do here? Since I know how to handle this exception, I would stop the execution here and return an error message to the caller. I approach errors and exceptions differently in my code. Another one I've had to deal with recently is hitting my usage limits with Cloudflare Email Sending (429). Nothing I could do about this, so I just decided to log the exception and fail fast.

• 
  Mehul Mohan (@mehulmpt) reported 6 minutes ago

  youtube is not going down every week because ai video slop is accelerating? cloudflare is not going down every week with so much bot traffic increase? i used to feel for github because of massive AI workload spike they have to handle, but it has been a while now. they had more than enough time now. this is a pure management, attention to detail, or talent issue at this point over github. disappointing

• 
  .null. (@Blacktrace_) reported 8 minutes ago

  @Cloudflare effectively putting the “bouncer” at the network layer rather than asking the agent to behave itself.

• 
  Shayan Nadeem (@shayannadeem123) reported 8 minutes ago

  @Taniyatweets_ GitHub but most website servers are behind cloudflare if that goes half of internet will go down.

• 
  Cortney Fletcher (@cfletcher24) reported 8 minutes ago

  @Cloudflare I can’t login to my account and the password reset isn’t working. And nobody is answering support. How can I find a support agent? My entire business is down right now.

• 
  Hero.S (@aka_ssy) reported 12 minutes ago

  Moving nameservers should be like changing the lock on your front door, not inviting the locksmith to install a little camera in the hallway. Cloudflare silently adding analytics JS to an HTML-only site is exactly the kind of “helpful default” that turns trust into a support ticket.

• 
  Nate Brown (@ntbrown01) reported 16 minutes ago

  @zackslab @fuzziphy I use Gitea (look/feel similar to GitHub) but exact same idea. I setup Cloudflare ZeroTrust, the agent runs on my router so I can be anywhere, flip this switch, and it’s like I’m at home on my network. I think Tailscale generally runs on a single system. But I want access to my whole network like I’m there.

• 
  Rahul Karajgikar (@Road_Kill11) reported 18 minutes ago

  most agent harnesses are not built to run in the cloud. this is why we rebuilt dexto from the ground up. we had a coding agent CLI similar to claude code, dexto a long time ago, but we were more interesting in making agents run in the cloud on a persistent computer. we started off by running dexto inside a cloud sandbox (e2b). this had some problems: 1. the runtime is heavily coupled to the sandbox. any sandbox outages or issues would kill the agent and were very hard to recover from. 2. credentials. we needed to give dexto access to many external integrations and tools. but we also needed to ensure that credentials weren't readable by the agent. we solved this by making an egress proxy (adds credentials on top of outbound requests from the sandbox so agents never see them). but this was painful and for every external integration we needed to maintain some code 3. price. we realized that for many of our use-cases (content creation, research, monitoring automations), the agent actually didn't need to read or write files. this pushed us to think that maybe it would be better if dexto ran OUTSIDE the sandbox, but could use the sandbox whenever it needed to write a file or run some command on the computer. while researching how agents could run outside the sandbox, i was very impressed with cloudflare's primitives. durable objects and workflows seemed extremely promising for what we needed and had good pricing. so i tried to deploy dexto on cloudflare. while doing this, i realized that our entire harness was designed assuming the agent would be running on a persistent machine and stored everything in-memory. this wouldn't work in a serverless environment. and interestingly, pretty much EVERY open-source harness you see (codex, opencode, etc.) has this same problem. so we redesigned our core primitives from the ground up, with the goal that these primitives should be able to power a local cli, webui, or desktop app, but ALSO be able to run on any cloud environment (cloudflare, railway, vercel, etc.). we changed every single primitive we had in dexto/core (storage, approvals, system prompt, skills, compaction logic, model registry, tool implementations) to be injectable from consumers so that they could define exactly how they needed dexto to run based on the required environment. while core just defined the shapes and the agent loop logic. now we're running dexto cloud fully outside the sandbox on cloudflare's stack, which has some amazing advantages: 1. dexto gets a persistent cloud computer with a browser and a terminal - but it only uses it when it needs to. this means that for requests that don't need a sandbox (like generating a video, or searching the web), our users spend $0 in compute costs. 2. the compute is now replaceable. it's not really an issue if we the sandbox has issues, or gets sandbox gets corrupted. we just replace the sandbox and move on. and only file writes/reads really are affected. 3. the agent loop is fully durable and resumable at any point, and don't cost extra while waiting. this means that dexto can wait 365 days for an approval, costing you no extra money. it also means if dexto fails something at any point, it can easily resume from the last safe point instead of having to re-run everything from scratch again. 4. credentials are easy! since the agent runs on a serverless environment, we don't need an egress proxy for 99% of normal requests. credentials never live on the sandbox, agents just hit internal apis and the credentials are managed separately so agents never see them. now we only use an egress proxy for a few specific CLIs that run on the sandbox (gh cli) if you're interested to learn more about harness design, and maybe see some cool demos and feature releases, follow along!

• 
  Sajid Mehmood (@smehmood) reported 19 minutes ago

  @axat_div @ethan_breitk @binsquares Cloudflare, no. For the others, they all started on the hyperscalers, but several of them are moving off in a major way (e.g. Railway Metal, Modal uses a number of neoclouds, etc.) And even if the hyperscaler is the underlying compute, it's meaningful that they are losing the direct customer relationship

• 
  Artyom Shimanski (@a_shimanski) reported 19 minutes ago

  @hectorivand @Namecheap @Cloudflare that's basically what I want, half my inbox is stuff I never agreed to

• 
  Mohit (@mohitdotdev) reported 21 minutes ago

  Our in house headless CMS completed migrations from PlanetScale to Cloudflare D1. Things that stood out 1. There were queries which read over 700M rows on a daily basis. I need to look into this as this is too much. 500k requests reading these many rows just isn't correct. Possibly indexing issue. 2. The DB size isn't much but API read logs were being written to the same DB. I would switch to Cloudflare Analytics Engine to count reads. 3. API reads would be moved to KV cache. This way DB reads get eased and expected response time becomes snappy. I hope this can be launched as a headless CMS as internal usages is all thumbs-up.

• 
  Rick A.F. (@FoxRick01) reported 22 minutes ago

  @andreigec My agents connected to Cloudflare handle all this admin, it’s not an issue at all tbh

• 
  Philippe Martin (@PhyByte) reported 24 minutes ago

  Writing the code was never the hard part. Owning what happens after it ships is. @cursor_ai bringing in a team that built agents specifically for that post-deploy loop (with real production experience from Cloudflare, Twitch, etc.) feels like one of the more grounded moves in this space with everything that happening at @SpaceXAI with @grok

• 
  Corrosion Resistance 🏳️‍⚧️🇵🇸 𓅃 (@licd) reported 34 minutes ago

  @SQNG Ah, so you moved from Cloudflare DNS direct to a service that partners with Cloudflare to serve their DNS. Could certainly be CF’s fault (or Porkbun, based on their website quality wouldn’t trust them lol), but you effectively moved DNS from CF to CF

• 
  MasterBismuth (@MasterBismuth) reported 34 minutes ago

  ****'s sake. Just managed to get my account unlocked because now my Firefox on my desktop can't seem to pass the stupid ******* Cloudflare challenges I'm constantly inundated with. Gonna lock my account but I doubt it'll help. Nikita Bier left this site in a dogshit state.

• 
  hexmint ✌️ (@hexmint) reported 37 minutes ago

  @Cloudflare you really need to QA this Create Worker page. If I enter two letters in any of the text boxes here, the first letter is lost. And if there is a single letter there, I need to press backspace twice to delete it. Feels like a React issue idk

• 
  Vance Vicious (Real, living human being) 🤘😎🤘 (@VanceVicious) reported 37 minutes ago

  @XFreeze Unfortunately for me none of the browsers on my pc will let me get to Grok. It keeps saying I'm blocked thanks to Cloudflare. I have asked @Support about this so many times, I can no longer count, but have yet to get a response. Shocking right?

• 
  mert (@ux_self) reported 39 minutes ago

  The most cost-effective $5 purchase I've made for Cloudflare Workers, and I’ve never regretted it.

• 
  Elenkova ✨ (@Elenkova_dxb) reported 39 minutes ago

  66% of AI-search citations already route through Google AI Mode, not ChatGPT. Same week Cloudflare started enforcing MCP's new auth spec, CIMD replacing OAuth DCR. Search and the agent layer are both quietly consolidating around infra nobody's hyping. That's the real signal.

• 
  Antonio Gomes (@onchainantonio) reported 41 minutes ago

  @Googlecloud's recent post-quantum roadmap highlights just how complex this migration is. The most honest line in it is about hardware. Google says some physical replacement cycles will run past 2029. This is a company that owns its data centres, its silicon, its browser and its certificate authority. And it still cannot promise the hardware layer is done by the end of the decade. That is the whole migration in one sentence. Post-quantum is a systems programme, not an algorithm swap. Three things in the roadmap that make the point: → Key exchange was the easy part. Hybrid ML-KEM is already live on Google's endpoints. Encryption in transit moves fast because it only needs two parties to agree. → Signatures are the hard part. ML-DSA and SLH-DSA are too large for the existing WebPKI to carry, so Google is testing Merkle Tree Certificates with Chrome and Cloudflare. They did not swap the signature. They redesigned how trust is distributed. → Roots of trust are the slow part. Caliptra, TPM 2.0, OpenTitan for quantum-secure boot. Silicon has a shipping cycle. No software update shortens it. Now remove Google's advantages. No single owner of the stack. No control over the clients. Keys and signatures published permanently the day they were made. That is what a decentralised network is migrating from. It is why "we'll upgrade later" is a much bigger claim there than it sounds. The organisations with the most control are being the most careful about the timeline. Worth sitting with.

• 
  Dipesh Acharya (@isDipesh) reported 41 minutes ago

  @vaibhavshn @karishnu @CloudflareDev Sign in with Cloudflare and Google is broken. Could only sign in with Github.

• 
  RED BARN NEWS 🇺🇸 (@JasperRoads) reported 42 minutes ago

  1/2 Quick tip that can save your parents/grandparents a ton of headaches: Grandma and Grandad keep getting hit with phishing texts and links. Every few months the phone gets completely wrecked and needs a factory reset. Changing the DNS to Cloudflare Family (1.1.1.3 + 1.0.0.3) blocks known phishing, malware, and scam sites at the DNS level. When they click a bad link, the page just doesn’t load instead of infecting the phone. It’s free, doesn’t slow anything down, and works quietly in the background. Not 100% perfect, but it stops a huge chunk of the common phishing junk older people get targeted with.

• 
  Adeilson Brito (@adeilsonrbrito) reported 43 minutes ago

  Cloudflare released one of the most consequential agent-security capabilities I've seen recently: its Gateway can now identify Model Context Protocol traffic at the network layer, surface previously invisible "shadow MCP" connections, distinguish direct MCP traffic from approved Portal-mediated traffic, and enforce policies that block MCP calls which bypass the governed route. There's an important architectural detail behind this. The new MCP 2026-07-28 specification exposes protocol and operation information directly in HTTP headers — "MCP-Protocol-Version", "Mcp-Method", and "Mcp-Name" — on every request, replacing the old connection-scoped handshake. Today, Gateway uses the protocol-version header to detect MCP traffic and enforce Portal-only routing at the network level. Cloudflare has indicated that tool-level policy — using "Mcp-Method" and "Mcp-Name" to govern individual tool calls — is coming next. Either way, the architectural shift is the same: infrastructure can increasingly identify, audit, and govern agent activity without depending on the agent itself to behave correctly. Cloudflare describes two distinct enterprise problems. Shadow MCP occurs when an employee connects an agent directly to an unapproved MCP server. Portal bypass happens when the server itself is approved but the user connects directly to it, skipping identity controls, DLP, curated tool catalogs, and tool-level audit trails. This is the key point: Agent security is beginning to look less like prompt security and more like Zero Trust. As agents become actors inside enterprise systems, security will increasingly depend not only on what we instruct agents to do, but on what the infrastructure allows them to do.

• 
  Sheik Althaf (@SheikAlthafDev) reported 43 minutes ago

  Shipped an LLM feature. Invoice arrived. Still couldn't tell you which customer spent what. This is the common case in 2026, not the exception. Most teams I talk to don't have a clean answer. The fix isn't a dashboard. It's six fields tagged at the call site: user_id, customer_id, feature, deployment, prompt_version, agent_run_id. Provider logs strip context. Post-hoc parsing breaks. Three failure modes from real incidents: → Runaway agent loops — retry-on-malformed-output billing tokens until a guard fires → Retry storms — SDK retries on transient 503s that the app never sees but the bill catches → Context bloat — a longer system prompt moves cost-per-request before total spend looks wrong Pricing is the second problem and it's worse than people think: • Reasoning tokens → 30–40% undercount if you only count visible output • 128K context crossing → silently 2x's per-token price, no signal in the response • Same model, different prices per host (Kimi K2.5: $0.5/$2.8 Together, $0.6/$3 Fireworks) Tooling is catching up. Cloudflare AI Gateway shipped dollar-budget limits in June 2026, scoped per model/provider/attribute. Portkey open-sourced their 3,500-model pricing DB. Langfuse v4 exposes cost_details per observation, queryable by tag. FinOps Foundation 2025: AI spend management jumped from 31% to 63% of practitioners YoY — the largest one-year swing in the report. Tag at request time. Enforce at the gateway. Then you can answer the question. How is your team attributing token spend today?

• 
  Erik Langille (@eriklangille) reported 51 minutes ago

  @BraydenWilmoth can you fix the two terrible iOS apps cloudflare already has? (1.1.1.1 and Cloudflare One) Tried using Warp for CF Zero Trust and it kept erroring. Turns out I needed this other CF One app. Took like 2 hours of debugging Tailscale is way smoother

• 
  Pete Stewart (@pj_stew) reported 51 minutes ago

  We currently use Bunny for video streaming at Achene, which is great, as it's much cheaper that almost all other video streaming services, which in turn allows us to deliver our online teaching service to each teacher cheaper than other platforms. Currently this is the only part of Achene that isn't on Cloudflare. There is a potential other option, which would be to create a video streaming service using R2, Cloudflare Containers for compression, and Workers for streaming and encryption... From a running cost point of view, Bunny is slightly cheaper for storage, but more expensive for bandwidth. There's a obvious cost to building our own streaming system, and it's not exactly a simple task, however, there are potential some significant wins if we did. One potential win, which I haven't tested, but feel would be the case, is an improvement to the UX for all users when uploading videos, as I feel it would be a quicker and smother service with it all on Cloudflare, and I could add some encryption and security that is either not available on Bunny, or is an additional paid service they offer that would make the service too expensive for each user. Some other tests I have done have suggested this would be the case. It would also mean that users could have more bandwidth for the same price, which would be beneficial to all that get lots of users. It's still a trade off, though, as you would get more bandwidth, but slightly less storage. Currently the cheapest tier on Achene is £5/$6 and give you 50GB Storage / 50GB Bandwidth. 1:1 ratio. With Cloudflare, if I were to reduce the storage by 10%, I could increase the bandwidth by 10x. I.e. 45GB Storage / 500GB Bandwidth. I would need to test this properly to be sure, but it would be something along those lines... Here's the questions... Do you feel it's worth while me testing this out as a viable option for video streaming? If so, should I do it now, before giving the first users access to Achene? Or should I continue with getting the MVP live, and put this to a later stage of development?

• 
  Fajar Sentosa (@fajarsentosabuz) reported 55 minutes ago

  @Cloudflare The hard part is whether access follows the agent's identity across the MCP connection, not just the employee's login.

• 
  Pils10 (@PilsZehn) reported 59 minutes ago

  @Nishanttt05 Namecheap, but I like Cloudflare as well... Namecheaps support is tier 1 though..