The mempool accelerator, explained: cost, pools and a live quote

Two searches bring people here. Mempool accelerator, meaning the service mempool.space runs, and txid accelerator, meaning any service where you paste a transaction ID and pay. They are the same thing in 2026, because the mempool.space Accelerator is the only large service built that way, by the people who run the explorer you were already looking at. Everything below was measured on 10 October 2026 from its public API, its public acceleration records, and the public mining pool statistics for the thirty days before.

What it is

The mempool.space Accelerator is an out of band fee service. You do not sign a new transaction and you do not raise the on chain fee. You pay mempool.space, and mempool.space tells its mining pool partners to treat your transaction as if it paid a higher rate. Its own FAQ gives the example: accelerate a 1 sat/vB transaction to a 20 sat/vB target, and partner pools will build blocks as though it paid 20 sat/vB.

It launched in July 2024 and was the first accelerator to publish every acceleration it takes, with the extra fee shown on the block page where the transaction was mined. That transparency is why we can measure it at all, and why it is the one paid accelerator we can describe with numbers rather than adjectives.

How it differs from RBF and CPFP

The FAQ is unusually direct about this and worth quoting in spirit: you should always use RBF or CPFP if you can. Both raise the real fee, both reach every pool on the network, and both cost only the shortfall. The accelerator is for the cases where neither works: a wallet with no bump button, a receiver with no spendable output, a transaction already purged from most mempools, or a long chain of unconfirmed ancestors that pins it. In exchange you accept a trusted third party in place of the peer to peer network, and a price that is a multiple of the shortfall.

The fifteen partner pools, with their share of blocks

Every active acceleration in the public feed lists the pool IDs it was sent to. We matched the fifteen IDs against the mining pool statistics for the thirty days to 10 October 2026, 4,414 blocks in all.

The share that is missing is just as informative. ViaBTC, at 9.4 percent the fourth largest pool, is not a partner; it runs its own accelerator and keeps its block space for that. SECPOOL at 4.1 percent and a tail of small pools make up the rest. So an accelerated transaction is invisible to roughly one block in five, and the expected wait is about 1.2 blocks, which matches what the records show below.

A live quote on a real stuck transaction

mempool.space exposes its estimate endpoint publicly, so we asked it to price a genuine transaction from the mempool that morning: 125 virtual bytes paying 13 sats, an effective 0.10 sat/vB, exactly the relay floor. Here is what came back.

Two things to take from that. The engine is honest: it computed a 1,469 sat shortfall for a thirteen deep package, which is a hard calculation most wallets will not even attempt. The shop front is less so: you cannot buy 1,469 sats of acceleration. You buy 7,000, and the difference is the service fee the FAQ mentions without ever stating a figure. For a simpler transaction we priced in the paid reviews, the calculation was 8,834 sats and the options started at 20,000. The pattern holds: expect the cheapest option to be several times the computed cost.

What thirty days of public records show

Every acceleration is published. We pulled the last thirty days on 10 October 2026 and measured them.

The numbers behind those tiles:

• 80 percent of the accelerated transactions were already paying 1 sat/vB or more when the buyer paid, and all of them were at 0.5 sat/vB or above. At a 1 sat/vB next block median, most of these transactions were already next block material, and the buyers paid for a queue position they held.
• The median extra fee was 2,500 sats, with a quarter of buyers paying 2,000 or less and a tenth paying 8,000 or more. The largest was 12,500 sats.
• The median resulting rate was 14.8 sat/vB, about fifteen times the rate the next block needed. Nobody needs fifteen times; the fixed purchase sizes make it unavoidable.
• Median wait: 16 minutes. Which is one to two blocks, which is what any transaction paying 1 sat/vB gets for free.
• Foundry USA mined nine of the fifteen, in line with its 25.5 percent share and its position as the partner most likely to find the next block.

Lifetime, since April 2023, the service has taken 31,043 requests and completed 30,426, a 98 percent completion rate, with about 2.9 BTC paid to pools in total. Volume has fallen hard: the current month runs at about 16 a day, and the thirty days we sampled contained fifteen. In a 1 sat/vB fee market there is simply not much to accelerate. These figures are re-measured every morning from the same public feed and republished at /data/facts.json.

The rules you accept when you pay

• No changes after purchase. "Once requested, an acceleration cannot currently be modified or canceled by the user."
• Cancellation without refund in conflict cases. If the service detects a transaction in conflict with the accelerated one, such as a replacement, it cancels the acceleration "and no refund will be given." Do not bump a transaction you have paid to accelerate.
• Partner pools choose. The FAQ says partners "who choose to accept your acceleration request" prioritise it. The 81.8 percent is the ceiling, not a guarantee that every partner acts on every request.
• A trusted third party. The FAQ lists this as the limitation itself: a centralised service in place of the trustless network. True, and refreshingly said out loud.
• Pro tier. For enterprise users, a maximum bid is set and the fee is calculated after the fact as the amount that would have sufficed. Lower cost, but it needs a preloaded account.

Is a txid accelerator the right tool for you?

How it compares with the free relay on this site

They do different jobs, and it is worth being plain about that. Our relay pushes a transaction to twenty nodes and tells you what is wrong. It fixes a transaction that never propagated or was dropped, and it diagnoses everything else. It cannot make a pool mine a low fee transaction, and we say so on every verdict. The mempool accelerator does exactly that one thing, for a price, through pools that find four blocks in five. Use ours first, because most days it tells you there is nothing to pay for. Use theirs when the verdict says nobody can bump and the payment cannot wait.

The bottom line

Among paid accelerators the mempool.space Accelerator is the one to take seriously: real pools, named and measurable, a public record of every job, and a price engine that computes the actual shortfall. Its weaknesses are the rounding up to fixed purchase sizes, the no refund rule, and a customer base that, by its own records, mostly did not need it. Check the fee rate first. If it clears the floor, keep your money. If it does not and you can bump, bump. If you cannot, this is the service to compare against the fourteen day wait.