Why this TPS is lower than the one you have seen
Until Oct 21, 2026, most of Solana's transaction count was consensus votes, the messages validators send each other to agree on blocks: 77% of all transactions up to then, and none of it work done for anyone using the network. Alpenglow, the consensus protocol that took over that day, no longer sends votes as transactions, so a count that included them fell by about 62% overnight while usage carried on. This chart leaves votes out of every day, which keeps its line comparable across the switch. A total that reaches back past it, all-time or in a cross-chain table, still counts them.
Why so many transactions fail
A failed transaction here is usually not a broken one. Fees are low enough that arbitrage and liquidation bots fire the same trade many times over and accept that most attempts lose the race: the loser is included in a block, reverts, and pays anyway. Volume is not what moves it: 2025 was the busiest full year on this chart and still got 69.1% through, while 2024 ran on less traffic and managed 51.7%. Because every one of those attempts pays the base fee, failures are not lost revenue for the validator that included them — they are part of what it earns.
What a Solana transaction costs
The median transaction pays 0.0000053 SOL, about $0.0005 — that is 5,309 lamports, 6.2% above the flat base fee of 5,000 lamports per signature. The average is higher, 0.000034 SOL, because it takes in the priority fees paid to jump the queue during a launch or a liquidation cascade: a small share of transactions carrying a large share of everything paid. Half of each base fee is burned and half goes to the validator that produced the block; priority fees go to that validator in full.
What this changes when you pick a validator
Transaction fees came to 71.7% of what a validator operator earned over the past week, with issuance commission and MEV making up the rest — the validator revenue chart shows the split. That is why this chart is an input to your yield rather than a curiosity: most of what your operator has to work with is decided by how much the network transacted this week, not by how much SOL is staked. And the admission ticket every validator pays to take part in consensus, about 0.7 SOL a day, stays the same in a quiet week, when there are fewer fees to pay it from. The live validator ranking compares operators on commission and performance, and the guide to choosing a validator covers what else to weigh.
Stake with an operator that keeps commission at zero
Stake.Cake takes 0% commission, so what you earn does not depend on how busy the network happens to be.
FAQ
Does a failed Solana transaction still cost me?
Yes. The fee is taken from the payer before execution begins, so a transaction that is included in a block and then rejected has already paid. Inclusion is what the fee buys, not success. The amounts are tiny: the median transaction pays about $0.0005. And most failures are probably not users' transactions at all, but bots that submitted the same trade many times over and lost the race. The series counts transactions, not who sent them.
Why is Solana usually quoted at tens of thousands of TPS?
That is a theoretical ceiling measured under laboratory conditions, not traffic. This chart shows measured daily activity, and it leaves out the validator votes Solana used to record as transactions, which were most of the count and no work done for users. Counted that way, Solana runs at around 1,711 transactions per second.
Does a high failure rate hurt what I earn from staking?
Not the way it looks. A failed transaction still pays its fee, so failures are not revenue the validator missed — they are part of what it collected, and fees came to 71.7% of operator revenue over the past week. The rate matters to you as a sender, when your own transaction is queued against everyone else's, not as a delegator. What decides your yield is your validator's commission and how reliably it produces blocks.
Where does this data come from?
Every row is one day of on-chain Solana block data: transactions that executed successfully, transactions that were executed and rejected, consensus votes counted separately while they were transactions, active addresses, and the fees paid. The series starts in 2020 and is refreshed daily. The headline figures are seven-day averages, because a single day is noisy and the newest one is often still being finalised.