TPS

TPS (transactions per second) is the most popular blockchain performance metric: how many transactions a network can process in one second. The number looks simple, but in practice there is a lot of confusion around it: marketing’s “theoretical ceiling”, the lab’s “stress-test peak” and the real “everyday baseline” can differ by tens or hundreds of times.

How it is measured

To compare TPS fairly, you have to fix the transaction type and the conditions:

• A simple native-token transfer is the lightest operation, requiring minimal computation.
• A complex smart-contract call — DEX swap, liquidation, DAO vote — is tens of times heavier.
• Test conditions — a lab rig, a testnet or a real mainnet under regular load.

Honest comparisons between networks should state what was actually measured. Otherwise a “100,000 TPS” of one chain ends up compared with “15 TPS” of another while the underlying transaction types are completely different.

TPS in TON

TON was designed for a high performance ceiling thanks to its sharded architecture. Each shard processes transactions in parallel, and the masterchain coordinates them, finalising the state.

• Theoretical ceiling. With the sharding model fully deployed — hundreds of thousands and potentially millions of transactions per second. This is a marketing boundary: it is reachable only with ideal load distribution and full activation of shards.
• Stress tests. The TON team and outside researchers have demonstrated peak values of order hundreds of thousands of tps in testnets over short intervals under synthetic load.
• Real baseline. On a normal day the mainnet processes orders of magnitude less — dozens of tps, sometimes hundreds during massive airdrops and hype campaigns (NOT, DOGS, HMSTR distributions). This is not a network limit but a function of how much real demand there is.

Why cross-chain comparisons are tricky

Bitcoin runs at roughly seven tps, Ethereum at around fifteen — but those numbers also include complex smart contracts, not pure transfers. Solana and BSC claim thousands of tps, but their baselines are also mostly simple transactions. Comparisons should be careful: same load type, same conditions, ideally an independent benchmark and not numbers from a landing page.

What matters for the user

• TPS is about infrastructure ceiling, not about the speed of an individual user transaction. Confirmation speed depends on block time and finality, not on TPS.
• During mass airdrops and hyped events the load can approach shard limits; wallets sometimes show delays in message confirmation.
• For everyday tasks — transfers, jetton operations, NFT mints — TON’s throughput is more than enough at the current state.

A high headline TPS is a nice property to have, but what it solves is not user experience — it is the network’s ability to grow when millions of active users actually show up.