Tezos activated Nairobi in its 14th protocol upgrade

Tezos activated its Nairobi protocol amendment at block 3,760,129 on June 24, 2023, at 00:07:10 UTC. The activation placed a new economic protocol into operation across the main network, changing transaction gas accounting, Smart Rollup functionality and parts of consensus-message handling.

Nairobi was Tezos’ 14th protocol upgrade. Its activation mattered because it converted code approved through the network’s amendment process into the rules applied by participating nodes. Rather than launching a separate chain or asking users to migrate assets, Tezos replaced the preceding Mumbai protocol at a designated block.

More capacity, with important limits

The most prominent claim attached to Nairobi was an increase of as much as eight times in transactions per second for certain manager operations. That figure came from experiments conducted by the protocol’s developers with a benchmarking tool distributed alongside Tezos and Octez. It was not a measurement of sustained production traffic on June 24.

The change made signature-verification gas charges more precise. Under Mumbai, manager operations paid a flat signature-checking charge that could overstate the required resources. Nairobi made the cost depend on the signature scheme and signed payload length, charged it only when a signature was checked and avoided repeating it across an operation batch.

Developers said this permitted roughly eight times as many basic transfers between accounts using the less expensive ed25519 and secp256k1 curves. Smart-contract calls were expected to realize a smaller gain, while operations using the more computationally demanding p256 curve received no comparable increase. The headline figure therefore was neither universal capacity nor a guaranteed mainnet rate.

Smart Rollups learned about protocol changes

Nairobi also extended the Smart Rollup system introduced under earlier Tezos upgrades. A new `Protocol_migration` internal message notified rollup kernels when the layer-one protocol changed. The mechanism allowed a kernel to identify an activation and begin using compatible functionality without relying entirely on an external signal.

The upgrade revised the WebAssembly proof-generating virtual machine to version `2.0.0-r1`. Existing Smart Rollups received that revision automatically at Nairobi’s first block, although some new capabilities still required kernel changes. The revision increased the virtual machine’s stack-size limit and added the `store_create` and `store_delete_value` host functions.

Typed transaction batches were another defensive improvement. Nairobi allowed a rollup kernel to specify the types of transactions sent through its outbox for execution on layer one. Developer documentation described the change as protection against ambiguous expressions that could otherwise expose a vulnerable rollup to unintended ticket withdrawals. It did not imply that such an exploit had occurred.

Consensus handling was refined

The amendment also relaxed how the node mempool accepted and propagated certain pre-attestations. Developers expected this to let votes for newly validated blocks travel before full block application completed, helping the network reach consensus earlier. “Endorsements” also began a terminology transition to “attestations,” reflecting that validators were attesting to a valid block rather than approving every transaction it contained.

These were protocol-level engineering changes, not evidence of immediate adoption, higher token value or additional application demand. The reviewed records establish that Nairobi activated and describe its implemented features. They do not establish the realized throughput of the live network after activation, the number of applications using the new rollup functions or a market-price response attributable to the upgrade.

The defensible June 24 conclusion was therefore narrower than the promotional headline: Tezos completed another coordinated mainnet amendment and expanded its scaling infrastructure, while the practical gains still depended on operation type, cryptographic curve, application behavior and actual network demand.