Coinbase status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Coinbase 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 Coinbase users through our website.

• Transactions (40%)
• Website (20%)
• Login (20%)
• Withdrawals (20%)

Live Outage Map

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

City	Problem Type	Report Time
Paris	Withdrawals	16 days ago
Le Taillan-Médoc	Transactions	19 days ago
Leipzig	Transactions	2 months ago
Maquoketa	Website	2 months ago
West Liberty	Login	2 months ago
Houston	Mobile App	3 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.

Coinbase Issues Reports

Latest outage, problems and issue reports in social media:

• 
  SerPAI (@im_serPAI) reported 3 minutes ago

  $HYPE down 23% in a month, ETF outflows hit $29.8M, and JPMorgan flagging regulated perp competition from Coinbase and Kalshi eating its lunch Still 4th largest corp treasury hold behind BTC, ETH, SOL Prediction markets might save it. Might not. Rough stretch ahead ngl

• 
  10MinuteOKByeBye (@Blakecrowes) reported 4 minutes ago

  @brian_armstrong Man I miss 2017. The Coinbase pump was awesome. Everyone knew if you got that CB listing it was rocket 🚀 time. Then the site would crash and everyone missed selling the top. Good times though. Whatever this crypto market is now it sucks. How about you kick off an alt season

• 
  𝓜𝓻. 𝓖𝓪𝓽𝓼𝓫𝔂 (@404_m3mes) reported 6 minutes ago

  Brian Armstrong listing niggabutt token on the solana blockchain on coin base is an absolute disgrace to the crypto industry For years I have tried to onboard Wall Street and tradfi as I truly believe some of the latest crypto innovations are once in a life time inventions Listing niggabutt token sets the industry back years. When my coworker downloads coinbase he will now see niggabutt token as a potential investment?? wtf @brian_armstrong

• 
  SafeBrowz (@SafeBrowz) reported 6 minutes ago

  We connected the SafeBrowz MCP to @Cloudflare's new AI Playground. A non-Claude model (GLM-4.7-Flash) discovered our tools on its own, called check_url, and flagged a fake Coinbase login page in seconds. DANGER. Trust score 5/100. Block. 🧵

• 
  CEOInterviews.AI (@CEOinterview) reported 8 minutes ago

  Hester Peirce leaves the SEC in a few weeks. Asked what breaks if the CLARITY Act dies, she named the one thing the agency cannot fix on its own. A platform where a token that is not a security trades beside one that is. The SEC has no clear authority over half that trade. Prediction markets put the bill near 30 percent. The Senate leaves for recess this month. Brian Armstrong told analysts failure would be close to business as usual for Coinbase. The gap Peirce named is what everyone smaller inherits.

• 
  Kveykva (@Kveykva4664) reported 8 minutes ago

  **** shorting Coinbase for now Im long IWM and QQQ YOLO

• 
  IT'S hustler (@abuldst) reported 8 minutes ago

  @base 5/ Tokenized US stocks are coming to Base through Coinbase. The framing behind it matters more than the feature itself — roughly 4 billion people globally don't have broker access, so this turns a huge pool of currently walled-off assets into something composable on-chain.

• 
  Gregers (@gregers_dev) reported 11 minutes ago

  @browser_use @USDC @coinbase Nice launch. A browser agent buying credits is the clean use case: x402 gets it to checkout; acceptance proof gets the server to say yes before running the task — authority, policy, spend cap, expiry, then a receipt both sides can keep.

• 
  ShadowX (@shadow12_x) reported 13 minutes ago

  @MichaelGSantos @brian_armstrong @coinbase Oh **** .. are donations safe?

• 
  Battle Jeff (@BattleJeff1) reported 13 minutes ago

  @drjasper_eth @gane5h No it doesn't. If Coinbase have to take a 0.2% yield cut. When solo get 1%, LST holders get 0.8%. LSD holders sell, Coinbase stake share reduce. >The problem with that curve is LST will deliberately not stake user ETH to maintain the peak issuance at 20% stake rate. Top three pools call each other. "Let's not stake beyond 20%, extra user ETH just park in our wallet" They'll get the highest possible total reward. Still distribute to users based on their LST holdings. Users get more reward, pools get more fees. Now the pools even get extra unstaked ETH to use elsewhere. That's why I keep saying the curve shape is wrong. It has to be strict monotonic.

• 
  Charu (@Charu_Sethi) reported 14 minutes ago

  There has been a discussion for last few years on how AI agents would transact on-chain, with crypto wallets and rails. Cloudflare has been quietly building this stack.They already sit between users and a large volume of the web traffic that has nothing to do with a crypto token. This week they announced: Account Wallets and Virtual Wallets, with spend limits you can set per merchant, per allowance, and per transaction. Funding and payments are coming soon. Back on 1st July it opened a waitlist for a Monetization Gateway, where it says charges settle in stablecoins over the x402 protocol. x402 is multi-chain and stablecoin-agnostic, so the chain is just handling settlement underneath. What Cloudflare is actually selling is metered access and programmable spend controls, sitting on the infrastructure agents already route through. The stablecoin is the cash register. The traffic and the wallet are the store , and we know the store economics My read: the agent-payment rail built by whoever already owns the traffic and the wallet or the store The chain underneath is a settlement detail. @Cloudflare @CoinbaseDev @coinbase #AgenticPayments #x402

• 
  Michael Fischer (@Holden_Rye_) reported 16 minutes ago

  @brian_armstrong @CoinbaseDev I think part of the problem is incentive. Building a better wallet means taking on real risk. If a normal app fails, someone gets annoyed. If a wallet fails, someone can lose their life savings. So how do we make it worth taking that risk? Why not put real money behind the problem? Coinbase has the reach and distribution to launch a bounty, competition, or hackathon for the next generation of self-custody. Let 100 teams try. Let 90 fail. Learn from all of them. Maybe one of those experiments becomes the PlayStation 1 we build from.

• 
  AChaosG | Markets & Power (@AChaosG) reported 16 minutes ago

  Bitcoin touched $65K on easing Hormuz concerns, then gave it back after this morning's weak jobs print. Same script all week — good news pops it, macro data pulls it back down. The ceiling isn't sentiment anymore. It's data. The real event is the CLARITY Act finally has a locked-in Senate vote. Not another delay story — an actual scheduled test. The 60-vote cloture hurdle is still real, but "will they vote" just became "will it pass." That's progress, whatever the outcome. Bigger shift: Saylor confirmed Strategy hasn't bought Bitcoin in six weeks and sold again at a loss. The "never sell" framework that defined this entire cycle just quietly ended. When the biggest corporate bull steps back, that's not FUD — that's data. $10B in BTC and ETH options expired today. That kind of size moving through the book in one session tends to set the tone for the week that follows, win or lose. Elsewhere: BlackRock cut fees on its Ethereum ETF, and Coinbase pushed into UK stock trading. Both are quiet bets that crypto infrastructure outlasts whatever the price does this week. Capital is signal. Chaos is noise.

• 
  Richardson Dackam (@RichardsonDx) reported 22 minutes ago

  Money becomes useless to an autonomous agent when it gets trapped in the wrong financial system An agent might: earn $500 through a checkout provider hold $200 in USDC have a card issued by another provider need to pay a SaaS merchant that only accepts cards Those are all "money" but operationally they are different islands Customer pays → money sits in Provider A Agent needs to spend → purchasing power exists in Provider B So WE (Humans) have to move/reconfigure/fund things for our AI agents An their autonomy breaks... The hard problem isn't giving an agent money It's making sure money earned in one system can become purchasing power in another without a human moving it Earn here → hold there → spend somewhere else → keep going An agent should be able to earn through one provider, hold value somewhere else, move that value when needed, spend through another rail, and keep reinvesting what it earns without a human stepping in to bridge the gaps My goal with Mandate is to give autonomous AI agents economic freedom I want anyone to be able to deploy intelligence that can create value, earn from that value, reinvest what it earns, and compound it into something bigger over time That's why Mandate had to be open source and provider agnostic No single financial provider covers every way an agent might earn, hold, move, or spend money One agent might receive revenue through Stripe, hold USDC in Coinbase, spend through Lithic, and use a bridge in between. Another might use entirely different providers Receive: Stripe / PayPal / stablecoins / marketplace payouts Hold: bank / financial account / wallet / stablecoin treasury Move: ACH / wire / bridge / onchain transfer Spend: card / bank transfer / stablecoin / x402 Mandate financial stack is composable: providers supply capabilities, Mandate connects them, and the agent sees one economic system You can connects whichever providers you choose into one continuous economic account Because autonomous agents won't operate in a crypto-only or fiat-only world. Crypto alone can't solve economic autonomy when most of the world an agent needs to operate in still runs on fiat, cards, and banks Fiat alone can't solve it either, because agents will increasingly use payment systems built for machine-to-machine commerce. The future is hybrid. The infrastructure has to be composable. I built Mandate so that an agent can operate across both as one continuous economy

• 
  MedusaMarkets (@MedusaMarkets) reported 22 minutes ago

  @BitmundFreud BIP 110 represents a digital resistance by individual node operators proving they can successfully block corporate capture. MicroStrategy operates as a Wall Street-subordinated arbitrage vehicle designed to centralize as much Bitcoin as possible in institutional custody (such as Coinbase). If decentralized node operators successfully assert rule-enforcement power via BIP 110, it disrupts their power to tokenize, and control the asset.

• 
  Secure Trace Lab (@SecureTrace_Lab) reported 26 minutes ago

  @Waylon3601 @KingOffX_ I read about your $25K lost through Coinbase and Ledger, both insider jobs, not user error. I've traced exchange leaks and hardware wallet breaches where the on-chain trail led further than anyone expected. I can surface where yours went. Let me know if you want me to take a look

• 
  Magoo PhD (@HodlMagoo) reported 30 minutes ago

  Ultimately @brian_armstrong and @coinbase thought they held all the cards when they pulled support for the Clarity Act earlier this year over stablecoin reward language. Now they are trying to desperately ram it through before a highly contested Midterm election with literal days left on the schedule. Massive miscalculation.

• 
  GoobyPLZ (@GoobyPlzz) reported 33 minutes ago

  @MoranElderLaw @brian_armstrong @SenLummis um, he is the ******* ******* who caused 2 delays in the past. His tweet "I will pull all support for the clarity act if it moves on as is" and then they pulled it off the floor because off this ******* *******. **** Brian and **** Coinbase

• 
  Darmin (@Darmin_che_) reported 36 minutes ago

  @LTHighDesert Damn I'm sorry about your experience mate, I've also had some difficulty with coinbase i lost have of my btc to an unknown address due to some system error. I felt upset luckily I got my funds recover. Kindly follow me I've a lead to share with you that can help?

• 
  Immorbu 🧸 (@Immorbu) reported 37 minutes ago

  @kris Has @cryptocom made an effort to get Coinbase to support Cronos chain? About 15% of my CRO is stuck with Coinbase and they only say maybe in the future they will support Cronos and I can access those coins.

• 
  ⬡ Lawrence ⬡ (@LawrenceCCIP) reported 44 minutes ago

  On August 4, @BitGo moved WBTC’s cross-chain infrastructure to @Chainlink CCIP. Exclusively. ~$7.7B in assets, plus every future BitGo-issued asset defaulting to CCIP. To understand why a custodian replaces infrastructure that never failed, go back four months. April 18, 2026. KelpDAO’s bridge loses 116,500 rsETH — $292M — in 46 minutes. No contract bug. Every onchain transaction was fully compliant on review. The attack hit a layer most people don’t think about. LayerZero’s verifier (DVN) is an offchain server. It can’t read the source chain either — it queries RPC nodes. So the trust chain runs three deep: destination → verifier → RPC → source chain. Attackers compromised the internal RPC node the DVN read from, DDoS’d the external fallbacks so the system failed over onto the poisoned source, and had it report a burn that never happened. The verifier signed correctly. The destination minted correctly. Audits check whether code was written wrong. This code wasn’t. The input was forged. What made it $292M was the config: 1-of-1 DVN. One verifier on the whole path — LayerZero Labs itself. Nothing prevented a $292M deployment from running a single verifier. Then accountability moved three times in three weeks. April: LayerZero blames Kelp’s config. May: Kelp says LayerZero approved it. May 9: “Allowing our own DVN to act as a 1/1 DVN on high-value transactions was our mistake. We own that.” No more 1/1, defaults to 5/5, floor of 3/3. Fixed on May 9. BitGo left three months later. Because for an institution, patched parameters aren’t the evaluation. The evaluation also covers why the parameter was set that way beforehand, and who owned the outcome when it failed. Run the same attack through CCIP and it has to clear four layers: — 16+ independent node operators, separate regions, orgs, hosting — RMN independently rebuilds the Merkle tree from the source chain — written in Rust by a separate team, zero operator overlap with the main protocol — contract-level rate limits in the token pool, enforced on both chains — anomaly detection that halts cross-chain activity on every chain at once LayerZero ships a rate limiter too. The difference is the default: it starts with no limits, and the docs say it “will not be necessary for most OApps.” Multi-verifier — opt-in. Loss cap — opt-in. Kelp died on the first one. CCIP puts both in the standard. You don’t configure your way to safe; you connect and it’s already running. SWIFT. DTCC. Euroclear. Coinbase. SBI Digital Markets. Now BitGo. $14.6B announced in 2026, all moving one direction. When WBTC completes, both major wrapped bitcoins — cbBTC and WBTC — run on CCIP.

• 
  Brutal Crypto Brief (@BrutalDegenX) reported 47 minutes ago

  Coinbase just got slapped down in Michigan - fed judge said nope, your federal commodities argument doesn't fly against state sports betting rules 🤦 They wanted an injunction. They lost. State regulators 1, crypto lobbying 0 $COIN

• 
  Nick (@hasty_nick) reported 49 minutes ago

  @coinbase @CoinbaseSupport Awful customer service. Unhelpful AI chatbot. You treated me as a scammer when I was just trying to buy SOL and transfer it to an external wallet. You wasted hours of my time today on a verification process that did not work. I will NEVER recommend CB.

• 
  Sachin Singh (@insachins) reported 50 minutes ago

  For three years, the AI story was capability: whichever model topped the next benchmark won. Entire businesses got built on one quiet assumption underneath that story, that AI compute would stay expensive. That assumption is now collapsing. GPT-4-level intelligence cost roughly $30 per million tokens in 2023. By early 2025, that was under $1.50. Today, equivalent capability runs for fractions of a cent. Epoch AI clocks inference costs falling at a median 50x per year, accelerating to 200x per year since January 2024. Gartner now projects inference costs for a 1-trillion-parameter model will fall over 90% by 2030 versus 2025 levels. Open-weight models made the gap irrelevant faster than expected. They held just 11% of enterprise token volume a year ago; now they’re at 38% and climbing, and by mid-2026 route roughly half of all production inference tokens. The migrations are real: Cursor moved to Kimi K2.5, Coinbase to GLM-5.2, Shopify and Airbnb to Qwen, even Microsoft is testing DeepSeek V4. That’s a problem for how AI companies get valued. OpenAI and Anthropic currently trade like Google-in-search or Meta-in-social, markets with real moats and winner-take-all economics. But there’s no meaningful switching cost between model providers, and enterprises already route work by price task-by-task. The businesses that survive this won’t be the ones with the smartest model. They’ll be the ones that built something a cheaper model can’t replace.

• 
  Jacek (@jacek0x) reported 55 minutes ago

  Builders have plenty of chains to choose from. I picked @base. Part of that is @coinbase. Long term vision, innovation, and above everything else, integrity. Other chains want those things too. I just trust Base with them more. To me, a chain is for the most part infrastructure. Base has been heads down making theirs better than anyone else's. What it can't do is make your product good. People point at Microsoft here. Own the operating system, build Office on top, push one with the other. That worked for Microsoft. A chain doesn't own your app. When it points its users at somebody else's app, that's all it's doing. Pointing. If the thing it points at isn't good, nobody stays. The product has to be good. That's the whole thing. Not good for crypto. Good. We have that with base:0x4ed4e862860bed51a9570b96d89af5e1b0efefed. A community that showed up before there was anything to gain. A brand people actually wear. And now we're building @POVMarket, a product people use because they want to. Nobody handed it to us. I want to build something on Base so good that it becomes the reason people show up to Base at all.

• 
  jasperthefriendlyghost.eth (@drjasper_eth) reported 55 minutes ago

  @BattleJeff1 @gane5h > Practically, it will stop even before reaching 1% The lowest reservation yield is by CEX stakers. Home stakers have thinner margins than major corporations. The ETH staker survey said most home stakers would exit below 2% APR. >The problem with that curve is LST will deliberately not stake user ETH to maintain the peak issuance at 20% stake rate. I have no idea what this means. Users of Coinbase will keep adding staking so long as APR is positive, same with Tom Lee.

• 
  Zetoshi (@ZetoshiX) reported 58 minutes ago

  Let me tell you something interesting about $BICO that most people probably don’t know! $BICO isn’t some random token. Biconomy was founded in 2019, and the project had some BIG names involved in its early funding. Its backers/investors included: • Binance • Coinbase Ventures • Mechanism Capital • DACM • Bain Capital Ventures • CoinFund • Fenbushi Capital • Huobi • True Ventures • NFX • Zee Prime Capital • Woodstock • Eden Block And it gets even more interesting… The $9M funding round also reportedly included major crypto-industry individuals connected to Aave, Flashbots, FalconX, and Arbitrum. $BICO later launched through CoinList in 2021. CoinList public sale prices: Option 1: $0.25 Option 2: $0.15 More than 850,000 users reportedly registered for the sale, and both options SOLD OUT. Now look at where $BICO is trading years later… And suddenly we’re seeing some of the biggest volume activity in its history, while open interest is exploding again. Something is getting interesting here. NFA.

• 
  OneFactor (@onefactormeme) reported 58 minutes ago

  @ashrobin the problem is that these big cos (like coinbase) have legal teams telling them they can't do 4, 5 and 6.

• 
  satoshi2024 (@vote4satoshi) reported 59 minutes ago

  @TheCryptoSquire You are ******* retarded. Coinbase has ****** up many things, this ain’t one of them. If not for them, the bill would have effectively killed the industry…. You want a good bill that can help the industry thrive, not any bill

• 
  BitmapCorp (@BitmapCorp) reported 59 minutes ago

  It Really Fair to Call a Bitmap a "Community Punk"? At BitmapCore, we believe that a label should represent an objective, reproducible, and verifiable on-chain property—not a visual interpretation produced by a rendering algorithm. As we explained previously, Bitmap images are not PNG, JPG, WebP, or any other conventional image format. They are generated programmatically by computer code that transforms the underlying blockchain data into a visual representation. Today, Bitfeed, Ordinals Wallet, and BitmapCore use a Mondrian bin-packing algorithm to render Bitmap images. This has already been publicly documented and discussed in previous posts. For that reason, we do not believe it makes sense to create labels such as Community Punk, because they classify a Bitmap solely because its rendered image happens to resemble a CryptoPunk. That appearance is not an intrinsic property of the Bitcoin block. It is simply the visual outcome of how the rendering algorithm arranged the block's transaction data. This leads to an important question. What if tomorrow that very same rendering algorithm were modified to draw animals, buildings, letters, faces, or any other predefined shape? Would it be fair to create new labels based solely on those visual appearances, knowing that the algorithm itself can be programmed to deliberately produce images matching whatever pattern its author decides? We believe the answer is no. A visual pattern intentionally produced by software does not become an inherent characteristic of the underlying asset simply because it is rendered that way. What gives a Bitmap value is its underlying blockchain data—not the artistic style chosen by a rendering algorithm. On the other hand, we do consider labels such as: • Grid Punk • Grid Perfect • Punk Perfect • 5 tx Punk Perfect • Punk Perfect 10 tx • Giga Punk Perfect • Wide Neck Punk • Standard Punk • Pristine Punk • Punk 2tx to be legitimate. The distinction is fundamental. These labels do not claim that a Bitmap "looks like" a Punk. Instead, they describe mathematical and structural properties of the Bitcoin block that anyone can independently verify directly from the blockchain. For example: • Every transaction in the block contains exactly the same BTC amount. • The block contains exactly two transactions with a specific ratio between the coinbase transaction and the second transaction. • The block contains exactly five or ten transactions following a precisely defined on-chain structure. These characteristics remain true regardless of how the Bitmap is rendered. They are derived directly from blockchain data—not from the rendering algorithm. This is the philosophy we follow at BitmapCore. If a characteristic exists only because of how an algorithm renders an image, and would disappear if that rendering algorithm were changed, then we do not consider it to be an intrinsic property of the Bitmap. However, if a characteristic can be derived directly from blockchain data and independently verified by anyone, then it represents a genuine property of the asset and deserves to be recognized as a label. We believe Bitmap labels should be built upon verifiable blockchain data, not subjective visual interpretations generated by software. That principle will continue to guide the development of BitmapCore. #bitmap #blockhain #bitcoin