Solana halved its block time to 0.21 seconds, with no outage

01 · FasterSolana now makes a block every 0.21 seconds

Solana now records a new batch of trades about every 0.21 seconds. In early August each batch took about 0.42 seconds. The network halved that wait in four steps over seven weeks, and kept running the whole time.

Each batch is a block. Validators, the computers that run Solana, take turns making them, and the chain gives each block a fixed slice of time called a slot. A shorter slot means a swap or a payment can land sooner, and the whole chain moves forward in smaller steps.

Bitquery records swaps on most Solana exchanges block by block, so we compared the chain before and after each step. We looked at how many blocks it made and how much each one held, what ordinary wallets paid, what changed for stakers, and how the fastest trading bots kept up. Blocks came twice as often and the chain never stopped. The network did not get twice as busy, though, and the bots lost no speed.

02 · The stepsHow did Solana cut its block time?

The change is a formal update to Solana's rules, a proposal called SIMD-0525. Validators switched it on in four equal steps, taking the target from 0.4 seconds a slot down to 0.2. Each step began at the start of an epoch, a fixed run of slots that lasted about two days at the old speed. The developers could pause before the next step if too many blocks went missing. The Solana Foundation's upgrade page gives the dates and two reasons for the change: shorter waits for users, and less time for any one validator to decide which transactions get in.

Our records show each step on the day it was due, and the public Solana node shows the last one after 9 October. Real slots ran a little slower than each target, as they did before the change.

03 · Smaller blocksIs Solana now twice as fast?

Blocks come twice as often, about 407,000 per day now. That alone does not double what the network can handle. With each step the proposal also cut the work a validator may pack into one block, in line with the shorter slot. The work allowed per second stayed about the same.

The trades show it. In early August a block held about 149 swaps on average. After the last step it held about 95. Swaps per day rose in between, from 30.7 million before the first step to 41.3 million in late September as memecoin trading picked up, then eased in October. Either way, the swaps now spread over twice as many blocks.

For someone sending a swap, the gain is in waiting time. A trade that just misses a block waits for the next one, and that wait is now about 0.2 seconds instead of 0.4.

04 · No outageDid anything break when Solana sped up?

No. In the 361 days we checked, up to 25 October, the chain never stopped. The longest gap between two blocks, once our own missing records are taken out, was under 5 seconds.

A skipped block is a slot where the validator whose turn it was produced nothing. Faster slots leave less time to finish a block, so this was the number the developers watched before each step. It went down. In random samples from the public Solana node, 0.23% of slots were skipped before the first step and 0.05% from 10 to 25 October.

Skipped blocks are only one way a busy chain shows strain. Failed swaps are another, and on Solana they come mostly from bots racing each other, as our study of failed swaps found. Their share moved with memecoin trading from day to day, before and after the change, so we make no claim about them here.

05 · FeesWhat does a Solana swap cost now?

The network fee on a typical swap stayed at about a tenth of a cent. Every Solana transaction pays a small base fee, plus a priority fee, a tip to jump the queue, if the sender adds one. For ordinary wallets, the ones that made 20 or fewer swaps that day, the typical swap stayed at that level on every day we sampled, from last November to this October.

The average is higher, because some wallets pay large priority fees to get ahead. It fell from 2.7 cents in November to about 1 cent on the October days we checked. Counted in SOL, Solana's own coin, the average about halved, and the price of SOL fell over the same months.

The network takes in less from each block. Swap transactions paid about 4,000 SOL per day in fees in early August and under 3,500 per day in October, spread over twice as many blocks. Each block now brings in less than half as much from swaps.

For most traders the network fee is still the small part. Trading apps often charge 1% of each trade, and our study of Solana app fees and Jito tips found those cost users far more than the network does.

06 · StakingWhat changed for Solana stakers?

Stakers got the plainest change. Staking means locking SOL with a validator to help run the network, in return for a share of newly made SOL. Solana pays those rewards once per epoch, and an epoch is a fixed number of slots. With slots half as long, epochs are too: one took 51 hours at the start of August and 26 hours in mid-October.

So rewards now land about once a day instead of about every two days. The yearly reward rate stays where it was, because the proposal also raised the number of slots the network counts in a year to match. Unstaked SOL frees up at the end of an epoch too, so a holder who stops staking now waits about a day at most instead of two, unless a lot of stake leaves at once.

07 · The botsDid faster blocks slow down the bots?

Snipers are bots that buy a new memecoin in its first moments, and in seconds they kept their pace. On Pump.fun, the site where most new Solana memecoins start, outside wallets bought into 42% of new coins within the coin's very first block in August. In late September and early October the share was 32%.

The shorter block explains the drop. Counted in time, about 53% of new coins got an outside buy within roughly 1.3 seconds before the change, and 52% after. The first block now closes sooner, and the bots arrive when they always did. Our study of Pump.fun snipers shows who those first buyers are.

Sandwich attacks did not fade either. In a sandwich, a bot buys just before someone else's trade and sells just after it, at the price that trade pushed up. By the daily counts behind our study of Solana sandwich attacks, bots made about 30 for every 10,000 Pump.fun swaps in the weeks before the first step. The rate dipped to 21 in early September, then rose to 33 after the third step, so the speed-up made no lasting difference.

By our rule, an attack has to fit inside three validators' turns, 12 slots in a row. That window shrank from about 5.0 seconds to 2.5 seconds, and the bots fit inside it just the same.

08 · Other chainsHow fast are Solana's blocks next to other chains?

Of the eight chains we compared, only Robinhood Chain makes blocks faster. In our own block records, Ethereum makes a block every 12 seconds and Base every 2 seconds. BNB Chain comes closest to Solana, and Robinhood Chain makes one every 0.10 seconds.

Block time is one measure among several. A block is not the same size on every chain, and the time until a block comes is not the time until a trade is final. Base, Optimism and Robinhood Chain are layer-2 networks, which run on top of another chain. On each of them a sequencer, one company's server, puts the transactions in order. On Solana, validators take turns.

09 · BuildersWhat do Solana apps need to change?

Any code that counts time in slots needs a look. A signed Solana transaction expires after 150 blocks, which used to take about 60 seconds and now takes about 30. A wallet that signs and waits, or a payment signed offline, has less time to send it. Code that turns slots into seconds by assuming 0.4 seconds a slot now gets every duration about twice as long as it should be.

Data grows too. Twice the blocks per day means twice the block records to store and read. The Solana data behind this story streams live through Bitquery's Solana streams, and the full history is also sold as files on the Bitquery Data Store.

10 · The upshotWhat did faster blocks change?

For someone sending a swap, the next block now comes twice as soon and the network fee stayed where it was, while stakers get paid about once a day. Underneath, the chain makes twice as many blocks, each holding less, so it does about the same work per second as before.

The bots barely noticed. Snipers and sandwich bots work to a clock in seconds, and the shorter blocks did not move it. The risk in the change was a chain that stumbles when it makes blocks twice as often, and that did not happen in the 361 days we checked.

11 · MethodHow we measured

We used Bitquery's records of every successful swap on the Solana exchanges it covers, over 343 days to 7 October, and its live trade records for the rest of October. Block time for a day is the seconds between the day's first and last swap, divided by the slots between them. We checked those times against the public Solana node, which also gave us the October readings and the skipped-block samples.

A halt would show up as two blocks far apart in time but close in slot number, so we looked for any such gap over the whole period. Fees are the network fee of each swap transaction, checked against the public node. Sniping counts new Pump.fun coins by whether a wallet other than the first buyer bought in the coin's first block, or in its first 1.3 seconds. The sandwich rate takes the daily counts from our sandwich study and divides them by Pump.fun swaps.

FAQ

What is Solana's block time now?

About 0.21 seconds, or roughly five blocks per second, since the upgrade's last step on 9 October. In early August a block took about twice as long.

What is SIMD-0525?

A Solana improvement proposal that halved the target time per slot in four equal steps, from 0.4 seconds to 0.2. Validators switched the steps on between 21 August and 9 October, each at the start of an epoch.

Did Solana go down during the upgrade?

No. We found no network halt in the 361 days we checked, and skipped blocks in our samples fell after the change.

Are Solana fees cheaper after the block-time cut?

The typical network fee for an ordinary wallet's swap stayed at about a tenth of a cent. The average, which large priority fees pull up, fell from 2.7 cents last November to about 1 cent this October.

Does a shorter epoch change Solana staking rewards?

Rewards arrive more often, every 26 hours instead of every 51 hours. The yearly rate stays the same, because the upgrade also raised the number of slots counted in a year.

How does Solana's block time compare with Ethereum?

Ethereum makes a block every 12 seconds, against about 0.21 seconds on Solana. Block time is not the same as the time until a trade is final, and blocks differ in size between chains.