Learn · Updated November 18, 2026

What is Solana transaction v1?

Solana transaction v1 is the transaction format defined in SIMD-0385 and active on mainnet since epoch 1035. It puts a version byte of 129 first and the signatures last, uses fixed-size counts, carries fee and compute settings in a configuration mask instead of Compute Budget instructions, allows up to 4,096 bytes and has no address lookup tables.

Where it was specified#

The format comes from SIMD-0385, "Transaction V1 Format", whose status is Activated. Its stated goal is a layout that is cheaper to ingest and process and that frees developers from Compute Budget instructions and lookup tables. Agave implements it behind the feature gate enable_tx_v1.

Legacy and v0 remain valid alongside it; a block can contain all three.

Field differences from v0#

Fieldv0v1
Version marker0x80 before the message headerbyte 129 (0x81) at the very start
Signaturesfirst, after a compact-u16 countlast, after the whole message; their count is the header's required-signature count
Countscompact-u16 (1 to 3 bytes)fixed one-byte instruction and address counts, two-byte data lengths
Recent blockhash32 bytes after the keyssame meaning, now called the lifetime specifier, earlier in the layout
Fees and computeCompute Budget instructionsa configuration mask and its values
Instruction layouteach instruction self-containedall instruction headers first, then all account indexes and data
Address lookup tablessupportednot supported

What is inside a Solana transaction describes the legacy and v0 layout these rows compare against.

Size and count limits#

LimitLegacy and v0v1 under SIMD-0385
Transaction size1,232 bytes4,096 bytes
Signatures1212
Accounts6464 (SIMD-0596, still a draft, proposes 96)
Instructions6464

The larger size is already in use. In two consecutive mainnet blocks on 18 November 2026, the largest of 272 v1 transactions was 2,805 bytes, more than twice what a legacy or v0 transaction may be, and one used all 64 addresses.

The configuration mask#

After the header comes a four-byte mask. Each set bit adds four bytes of configuration values, in bit order:

  • bits 0 and 1 together: the priority fee, as a total in lamports (both bits must be set);
  • bit 2: the compute unit limit;
  • bit 3: the loaded accounts data size limit;
  • bit 4: the requested heap size.

Unset fields take their minimums, not the old defaults: no priority fee, a compute unit limit of zero and a loaded-data limit of zero. Compute Budget instructions inside a v1 transaction are ignored for configuration, though they still run as no-ops. The priority fee being a total, not a price per compute unit, is the change most likely to break fee analytics; compute budget instructions covers the older model.

In the same two blocks, every v1 transaction set a compute unit limit and a loaded-data limit, 245 of 272 set a priority fee, 15 asked for a larger heap, and 8 still contained Compute Budget instructions that the runtime ignored.

Activation status, as of 18 November 2026#

The enable_tx_v1 feature account on mainnet records activation at slot 447,120,000, the first slot of epoch 1035, whose block time was 15 September 2026, 01:04 UTC. RPC methods return v1 transactions only when the caller passes maxSupportedTransactionVersion: 1. Of the 781 transactions in our two sampled blocks, 272 were v1.

What v1 changes for a shred decoder#

Entries carry v1 transactions in exactly this wire form. A decoder must check the first byte before anything else, because reading a v1 transaction as legacy takes its ID from the wrong bytes and loses its place in the entry. In exchange, v1 is easier in two ways: every account is in the transaction, so there are no lookup tables to resolve, and the fee is stated outright. Signatures as IDs shows where the ID sits in each format.

Questions about transaction v1#

Is transaction v1 live on mainnet?#

Yes. Its feature activated at slot 447,120,000, epoch 1035, and v1 transactions appear in ordinary blocks.

Do I need lookup tables for v1 transactions?#

No. The format has none; every account is listed in the transaction.

Is the v1 priority fee a price per compute unit?#

No. It is a total in lamports, set in the configuration mask.

What the format does not change#

v1 changes how a transaction is laid out, not what a raw transaction can tell you. It still holds only the sender's request: no success flag, logs, inner instructions or balance changes. The recent-blockhash rule for expiry is unchanged, and so is the meaning of the first signature as the transaction's ID.

Reading v1 in practice#

The Rust decoder and the TypeScript decoder both handle v1 next to legacy and v0, and shred parsing libraries compared lists which libraries do. To see v1 transactions in live traffic, decode a first transaction from raw shreds.

Sources#

Raw shreds on your own server

Ultra Low Latency Shreds is $200 and Standard Shreds is $100. You pay in crypto, tell us which IPv4 address and UDP port should receive the stream, and we wire it up by hand.