Every block explorer,
actually tested.

Why a whole site about block explorers

Almost everyone who touches crypto ends up on a block explorer within their first week, usually in a slightly panicked state. They have sent something and it has not arrived. They have been given a transaction ID by a support agent and told to check it. They have been sent a screenshot by someone on the internet who claims it proves a payment. The explorer is where all of those questions go to be answered, and almost nobody explains how to read what comes back.

The tools themselves are not the problem. Most of them are technically excellent, and several are free in a way that is genuinely remarkable given what they cost to run — a full Bitcoin node with a transaction index is a serious piece of infrastructure, and Blockstream hands out API access to it without so much as an email address. The problem is that these tools were built by engineers for engineers, and the interfaces assume you already know what a change output is, why the "amount" on a Bitcoin transaction is usually wrong, or what it means that a Base transaction is confirmed but not yet settled.

So this site does two things. It reviews the explorers properly, against criteria that are written down and applied the same way every time. And it explains the ledger underneath, in the specific places where the interface misleads you.

How we actually test them

Every explorer gets the same treatment. We paste in a fixed set of reference items — the genesis block, a transaction with an unusually large number of inputs, an address with a long history, a failed contract call, a pending transaction sitting in the mempool — and record what each explorer does with them. Anything that is wrong, missing or silently rounded goes in the notes.

Then we read the API documentation and actually call the endpoints. This matters more than it sounds, because published rate limits and real rate limits are frequently different numbers. It is also how we caught the shift that happened this year: several explorers that were genuinely keyless in 2024 now require registration, and a few of the "free" tiers have quietly become small enough that you would exhaust them during a single afternoon of development.

The privacy score is the one that produces the most argument. An explorer that logs every lookup, fingerprints the browser and shares data with a commercial analytics arm is doing something meaningfully different from one that serves static pages over an onion address. Both can be excellent at showing you a transaction. Only one of them keeps that lookup between you and the ledger. We do not pretend that difference is a matter of taste.

What a block explorer cannot do

This is the section people skip and then email us about, so it comes early. A block explorer is a window onto data that already exists. It cannot reverse a transaction, cancel one, or recover funds sent to the wrong address. It has no relationship with your wallet and no ability to act on your behalf. If a page or a person tells you that it can, you are being set up.

It also cannot tell you who owns an address. Analytics platforms such as Arkham attach names to wallets by combining on-chain patterns with off-chain information, and they are often right, but "often right" is not the same as authoritative. Attribution is inference. Treat a label as a hypothesis rather than a fact, particularly if you are about to make a decision based on it.

And it cannot show you what is not there. On Monero the amounts, senders and recipients are cryptographically hidden, so an explorer can confirm a transaction exists and essentially nothing else. On Litecoin's MWEB the same thing is true inside the extension block. An explorer showing you a blank where you expected a figure is usually working correctly.

The three things people get wrong

First, the amount. On any UTXO chain — Bitcoin, Litecoin, Dogecoin, Bitcoin Cash — a transaction spends whole outputs and sends the remainder back to itself as change. An explorer that reports "total output" will therefore show a number far larger than what was actually paid to anyone. Someone sending 0.01 BTC from a wallet holding 2 BTC produces a transaction with roughly 2 BTC of output, and screenshots of exactly this have been used to convince people that enormous sums were moved. Our field guide to reading a Bitcoin transaction works through a real example.

Second, confirmations. There is no universal number. One confirmation on Bitcoin represents about ten minutes and a great deal of accumulated work; one confirmation on a chain producing blocks every two seconds does not. Exchanges publish their own thresholds and those thresholds are the ones that govern when your deposit credits, regardless of what the explorer says.

Third, finality on rollups. A transaction on Base or Polygon shows as confirmed almost immediately, but it is not settled on Ethereum until its batch is posted and — for optimistic rollups — the challenge window has elapsed. Two different timestamps, two different meanings, one word on the interface.

When the documentation is wrong

Every API claim on this site comes from a request we made, dated, rather than from a pricing page. That sounds like pedantry until you run the requests, at which point the gap turns out to be wide and consistently in one direction.

Blockchair documents roughly thirty requests a minute on its keyless tier. Testing it on 17 September 2026 we were cut off after about ten — not throttled but blacklisted, HTTP 430, every endpoint including the free stats call that had worked a second earlier. That distinction decides an architecture. A throttle you back off from; a block you cannot, and if the requests come from one server rather than from each visitor's browser, a single burst takes the feature down for everyone.

Etherscan advertises around sixty networks under one V2 key and roughly ninety per cent chain coverage on the free tier. Both true, and both misleading: the percentage counts chains, not usage. A free key answered on Ethereum, Polygon, Arbitrum, Linea and Blast, and was refused outright on Base, BNB Smart Chain, Optimism, Gnosis and Avalanche — which is to say, on several of the busiest EVM networks in use.

Solana's public endpoint is the subtlest of the three. It answers a terminal and refuses a web page, returning 403 to any request carrying an Origin header. Test it with curl and you will write down that it works without a key, which is true and useless if your code runs in a browser. We test with the header set, because that is what the reader's browser will send.

None of this is difficult. It is about twenty minutes per service, and it is the whole difference between a review and a rewritten press release. Our methodology sets out the four checks we run and why each one catches something the others miss.