Solana activated its v1 transaction format on mainnet on Tuesday, raising the maximum size of a single transaction from 1,232 bytes to 4,096 bytes. The change is the first rewrite of the network’s transaction format since v0 introduced address lookup tables, and it clears the way for workloads that never fit inside one Solana transaction before.
The upgrade combines two Solana improvement documents. SIMD-0296 lifts the size ceiling, while SIMD-0385 defines the v1 message format that carries it. The format has been live on testnet since epoch 1025 on September 1, and developer shop Anza confirmed the September 9 mainnet date on August 29 after a week of incident-free operation on testnet.
Why the old limit existed
The 1,232-byte cap was an accident of network plumbing. Solana’s engineers sized early transactions conservatively around a 1,280-byte IPv6 maximum transmission unit. When QUIC became the default transaction ingestion protocol in 2022, the ceiling stopped making technical sense: QUIC’s specification, RFC 9000, sets no maximum stream size, so larger payloads can cross the network without fragmentation at the MTU layer. The limit stayed in place anyway, and for four years it shaped what developers could build.
The size increase matters most for cryptographic workloads. Zero-knowledge proofs used in Confidential Transfers under the Token Extensions standard produce payloads larger than 1,232 bytes. So do Winternitz one-time signatures, BLS signature schemes, and the nested multisig structures that institutions use for custody. Until now, those workloads either required multiple chained transactions or were simply impossible onchain.
What changes for builders
v1 is opt-in for senders. Legacy and v0 formats keep working exactly as before, so wallets and applications that do not need the extra room can ignore the change. Applications that do adopt it must set the compute unit limit and loaded accounts data size explicitly, because the defaults are now zero and unsigned transactions fail without them.
The new format replaces ComputeBudgetProgram instructions with a 32-bit TransactionConfigMask at a fixed offset in the message header. The mask encodes priority fees, compute unit limits, loaded accounts data size and heap size directly. Per-instruction limits hold at 200,000 compute units with a transaction-wide cap of 1.4 million.
There is a trade-off with address lookup tables. v1 transactions give up some of the account-count flexibility ALTs provide in exchange for the larger payload, and the Solana Foundation published guidance in August on how developers should weigh the two. Most applications will not need to choose: the majority of Solana traffic today fits comfortably in v0 with ALTs, and the new format targets the niche that did not fit at all.
Breaking change for infrastructure
Reading blocks is where the change bites. RPC providers, indexers and block explorers must handle v1 traffic the moment it appears on mainnet, and the Solana Foundation’s migration guide lists the failure modes: incorrect decoding of the new header, missing fee data, and silently dropped transactions. Anza requires providers to run Agave v4.2.2 or v4.3.0-beta.3 or later to process v1 correctly.
Minimum SDK versions are documented as well. Applications need @solana/kit 8.0.0 for reading and sending, web3.js 1.x at 1.99.0-beta.0 for read-only support, and the Rust crates at 4.2.x. Local testing has been available since August 24 through the Solana CLI v4.2 and Surfpool v1.5.
The activation lands during a quiet stretch for the asset. SOL trades near $103, down about 1% on the day, and weekly inflows into Solana ETFs slowed to $6.18 million last week from $153.87 million the week before, according to CoinEdition market data.
Adoption now depends on the ecosystem rather than the protocol. Confidential Transfers, institutional multisigs and post-quantum signature research are the headline use cases, but each needs application-level updates before anyone can send a v1 transaction. If indexers keep pace with mixed traffic over the coming days, the format rewrite will have done its job quietly, which is how infrastructure changes are supposed to land.
discussion