BTCPay vs BlockBee vs BitHide: Self-Hosted Compared

If you are comparing BTCPay Server, BlockBee and BitHide for a self-hosted crypto payment setup, the first thing to fix is the premise: only two of the three are self-hosted, and the one that is most often lumped in with the others is not.

BTCPay Server is open-source software you deploy and run yourself, and it never asks for a private key. It takes an extended public key and watches for payments. BitHide is also software you install on your own server, but its source is not published, and you reach a deployment through a sales conversation rather than a download. BlockBee is neither: it is a hosted API. You call it, it generates a receiving address, the customer pays that address, and BlockBee forwards the funds to the wallet you nominated. Nothing runs on your infrastructure.

That one distinction decides almost everything else: who can freeze a payment, what happens if the vendor disappears, and whether “self-hosted” buys you any auditability at all. Below is what each one actually does, checked against the vendors’ own pages on 7 September 2026, followed by the thresholds where self-hosting stops being worth it.

Where we stand

We build Aurpay, a hosted non-custodial payment gateway. That makes us an interested party in any “should you self-host?” question, so read the recommendation sections with the scepticism they deserve. We have also written out, in plain terms, the three situations where running your own stack beats paying us. Those situations are real and they are not rare. If your business is in one of them, self-hosting is the correct answer and you should take it.

The comparison

Every cell below links to the vendor page it came from. All were checked on 7 September 2026. Where a vendor does not publish a figure, the cell says so rather than guessing.

Dimension	BTCPay Server	BlockBee	BitHide
Architecture	Self-hosted software you deploy	Hosted API service. Nothing runs on your server	“A self-hosted, non-custodial solution installed on your server”
Where the payment first lands	An address derived from your own wallet	An address_in that BlockBee generates, then forwards to your address_out	An address on your own installation
Who holds the private key	You, and only you. “the software only needs your extended public key”	Not disclosed. BlockBee controls the intermediate address; its docs offer no way to supply your own extended public key	You. “You store your private keys and seed phrase, 100% under your control”
Source code	Public, MIT licensed	Not published (hosted service)	Not published. The only repository under its GitHub organisation is empty
Assets and networks	Bitcoin and Lightning in core. “Bitcoin is the only focus”; everything else runs on community plugins	“102 assets · 12 networks” on its asset page	BTC, LTC, ETH, TRX, BNB, USDT, USDC across ERC-20, TRC-20, BEP-20
Service fee	“No subscriptions, no transaction fees”	1.000% deposit / 0.500% payout at the entry tier, falling toward 0.25% on volume	Not published. The site says “Let’s talk”
Blockchain fees	Paid from your own wallet when you spend	Charged on top: “plus blockchain/network fees”	Paid by your installation; markets a Tron energy feature claiming to “reduce TRX costs by up to 50%”
Blockchain node	Yes, you run one	None	Not documented publicly
Minimum server spec	2 GB RAM, 80 GB storage with pruning enabled, plus more per extra chain	None	Not published
Realistic infra cost	$20/month at LunaNode’s cheapest plan that clears both minimums	$0	Your server plus an undisclosed licence
Upgrades and backups	Yours	Vendor’s	BitHide states it handles “deployment, configuration, updates”
Project activity	v2.4.4 released 7 Sep 2026; ~7,700 stars, 200+ contributors	Commercial, closed roadmap	Commercial, closed roadmap
Pricing transparency	Full. The software is free	Public calculator, tiers not enumerated	None public

The private key column is the whole argument

Three products, three genuinely different custody models, and the marketing language for all three uses the word “non-custodial.”

BTCPay Server holds nothing, by design. You give it an extended public key. From that it derives a fresh receiving address per invoice and watches the chain for a matching payment. It cannot spend, because it was never given the ability to. That is the strongest position of the three, and it holds because of how the software is built rather than because anyone promised it. If your BTCPay box were compromised tomorrow, the attacker would walk away with your transaction history, not your balance.

BitHide holds the keys on hardware you control. The security claim is credible in shape: keys on your server, seed phrase in your possession. What you cannot do is verify it. There is no published source, no public installation documentation, and no independent audit we could find. You are running a closed binary that generates and stores private keys, and the assurance that it does so correctly is the vendor’s word. That is a materially different trust model from BTCPay’s, even though both are described as self-hosted and non-custodial.

BlockBee sits in the middle, and it is the one people misread. The flow is documented plainly: BlockBee creates address_in, your customer pays it, BlockBee forwards to address_out. Funds settle to your wallet, which is why the “no custody” language is not dishonest. But between the customer’s send and the forward, the money sits at an address you did not generate and cannot spend from. There is no option in the documentation to supply your own extended public key. Whether that gap matters depends on your risk appetite; whether it exists is not a matter of opinion. It is the same distinction we walk through in our custodial versus non-custodial breakdown, applied to a service that markets itself on the non-custodial side of the line.

One practical consequence of the forwarding model: BlockBee’s per-asset minimums are enforced, and payments below them are lost. Its own asset page lists a 0.00008 BTC minimum for Bitcoin. If your average order value sits anywhere near those floors, that is a checkout-level problem you have to solve in your own code, not a footnote.

The costs nobody puts in the table

The standard comparison post says BTCPay is free and stops there. Two numbers complicate that.

The server is more expensive than the usual quote. BTCPay’s documented minimum is 2 GB RAM and 80 GB of storage with pruning enabled. Comparison articles routinely cite $8 to $14 a month for a suitable VPS. We checked LunaNode’s current pricing on 7 September 2026: the 2 GB plan ships 35 GB of storage, less than half the documented minimum. The cheapest plan clearing both figures is $20 a month, and BTCPay’s own docs add that you must “expand the storage according to that coin(s) blockchain size” if you run anything beyond Bitcoin. Call it $240 a year before you have spent a minute of your own time.

Chain fees are not symmetric, and the industry has them backwards. We measured settlement costs across our own transaction flow, and the result inverts the received wisdom. ERC-20 transfers landed around $0.02. TRC-20 transfers cost $2.11 to $4.36 over the same sampling window. Tron is widely described as the cheap rail for stablecoins; on our measurements it is roughly a hundred times more expensive per transfer than Ethereum right now. This is not an argument against Tron — customer wallets and habits matter more than a few dollars on a large order — but it does mean the “we saved you gas fees” line in any gateway’s marketing needs a chain attached before it means anything. Our fuller working is in the TRC-20 processing breakdown.

Both of these feed the same calculation. Self-hosting trades a percentage fee for a fixed cost plus your attention. The fixed cost is roughly $240 a year in hosting. The attention is the part that decides it.

Where the breakeven actually falls

Take a 1% hosted fee as the comparison point. Against $240 a year of hosting, you break even on cash at roughly $24,000 of annual crypto volume, or $2,000 a month. Below that, self-hosting costs you money outright. Above it, you start saving, and at $500,000 a year you are saving roughly $4,760.

Now price the maintenance. Running BTCPay means node upgrades, certificate renewals, database backups you have tested, monitoring that pages someone when the node stalls, and a plan for what happens when the person who set it up leaves. Two to four hours a month is a fair estimate for a stable single-chain deployment; add Lightning and channel liquidity management and it goes higher. At three hours a month and $75 an hour of loaded cost, that is $2,700 a year, which pushes the real breakeven past $290,000 of annual volume. Call it $24,000 a month.

So the honest thresholds:

• Under roughly $25,000 a month in crypto volume: self-hosting is a hobby, not a saving. The fee you avoid is smaller than the time you spend, and you have added a payment-critical server to your operational surface.
• $25,000 to $100,000 a month: genuinely close. It comes down to whether the maintenance lands on someone who is already administering servers for you. If yes, self-host. If it means hiring or learning, it does not pay.
• Above $100,000 a month: the arithmetic favours self-hosting clearly, and the question stops being cost and becomes reliability. Downtime on your own gateway is lost revenue you cannot escalate to anyone.

Three cases where self-hosting is simply correct

Independent of volume, some requirements can only be met by running the stack yourself.

You need the settlement logic to be yours. Split payments across parties, confirmation thresholds that vary by order value, settlement that has to reconcile against an internal ledger before release. Hosted gateways expose a fixed flow and you build around it. On BTCPay you change the flow. If your business model lives inside those rules, no amount of convenience compensates.

A regulator or counterparty requires funds never to touch third-party infrastructure. Some licensing regimes and enterprise procurement reviews treat any intermediate address as a control gap, regardless of how briefly funds rest there. This is where BlockBee’s forwarding step fails an audit that BTCPay passes, and it is much better to know that before the review than during it.

Bitcoin and Lightning are the business. BTCPay’s Lightning support is core rather than a plugin, and no hosted gateway comes close on node-level control, channel management or point-of-sale deployments. If Lightning is the product and not just one option at checkout, this is the tool.

The mirror image is also worth stating. Self-hosting is the wrong call when stablecoins across several chains are the bulk of your volume. BTCPay’s stated policy is that “BTCPay developers do not implement alternative coins on request”, and plugins that lose their maintainer get removed. Building a stablecoin business on a plugin whose continued existence depends on volunteer attention is a risk you are taking on purpose. Our head-to-head on Aurpay and BTCPay Server goes through that trade in detail.

How to choose in five minutes

Answer these in order and stop at the first yes.

1. Does a regulator, auditor or major counterparty require that funds never rest on third-party infrastructure? Self-host. BTCPay if you want auditable source; BitHide only if you are willing to accept an unauditable binary in exchange for its multi-chain coverage.
2. Is Bitcoin plus Lightning more than 80% of your crypto volume, and do you have someone who already runs servers? BTCPay Server.
3. Do you need custom settlement or split-payment logic that no hosted checkout exposes? BTCPay Server, and budget for engineering rather than just hosting.
4. Are you above roughly $100,000 a month with an ops function? Model both. The fee saving is real at that scale.
5. Everything else. A hosted gateway, and the only question left is whether it takes custody. Compare the hosted providers side by side on the key-control column, not the fee column — the fee difference between hosted providers is a fraction of a percent, and the custody difference is all of your money.

Where Aurpay fits

We are on the hosted side of this comparison, with one structural difference from BlockBee’s model. Payments settle directly to the merchant’s own wallet rather than passing through an address we generate. The fee is a flat 0.8% per transaction, with no contracts or banking details required to start. We support Bitcoin mainnet, Bitcoin Lightning, Ethereum and Tron — BTC, BTC over Lightning, ETH, USDT and USDC on ERC-20 and TRC-20, and DAI on ERC-20. That is the complete list. It is deliberately narrower than BlockBee’s 102 assets.

We do not offer a self-hosted build, and we are not going to pretend the hosted trade-off is free. You are paying 0.8% to not own a server. If your volume and your team make that a bad deal, BTCPay Server is a serious piece of software and the arithmetic above tells you where the line is.

If it is a good deal, see how the hosted options line up in our crypto payment gateway comparison, check the numbers against our gateway fee breakdown, or install Aurpay on Shopify or WooCommerce and start taking payments straight to your own wallet.