Introduction
The newly released version of Starknet – previously referred to by its numerical version name v0.14.0 and now known as ‘Grinta’ – marks a major milestone in Starknet’s decentralization journey by distributing the sequencer architecture. While Starknet previously operated with a single sequencer, it now has three independent sequencers running consensus and taking turns building blocks.
This version of Starknet also includes some important user-facing features, including subsecond pre-confirmations, a mempool, a fee market, and a standardized integration to paymaster, all of which are discussed below.
Strides toward Decentralization
The biggest change introduced by Starknet Grinta is a decentralized sequencing architecture consisting of three sequencers running consensus. Trust minimization is one Starknet’s core values and decentralized sequencing is essential to ensuring neutral transaction inclusion and ordering. Together with Staking v2, which was introduced in June 2025, this version of Starknet is a significant step toward putting the necessary technical infrastructure in place for the next phases of decentralization.
During this version, the operation and control of the sequencers is still centralized under StarkWare; however, according to Starknet’s decentralization roadmap, block validation and production are distributed between several nodes operated by StarkWare, with fully decentralized sequencing and proving to come in the future. Decentralization is a core requirement for a public blockchain, and for Starknet this will mean that protocol security is managed through transparent, community-driven mechanisms.
Additional Features
Pre-Confirmations
Starknet Grinta introduces incredibly fast, subsecond pre-confirmations, which means users receive the transaction status much earlier than block publication. In other words, pre-confirmation is a promise by the block proposer that a transaction is part of the proposal submitted to the consensus. The advantages of pre-confirmations for users are significantly reduced latency and the ability to act on transactions before L2 finality.
Mempool
Mempools are also being introduced to Starknet with Grinta. Mempools store transactions waiting to be added to a block. They are continuously updated by adding, holding, deleting, and replacing transactions that are received. Starknet’s sequencers’ mempools are connected in a p2p network, so they are synced about transactions within a fraction of a second when there is no congestion.
Fee Market
Another new feature focused on future-proofing Starknet is a fee market on l2_gas. Prior to Grinta, the price of l2 gas was dependent on Ethereum’s gas price and, as a result, also affected by fluctuations on Ethereum. In addition, gas fees covered the “marginal” onchain cost of verification. However, due to significant technological improvements on Starknet, including advances in its proving stack and the introduction of blobs, onchain costs have become almost negligible.
As Starknet moves toward decentralization, we are striving for a fee policy that incentivizes the participation of many sequencers. With this in mind, and together with feedback from the community, the minimum l2_gas base price is set at 3 gFri. This base price ensures that fees cover the operational costs when gas prices on Ethereum are low, while improving the scale and costs when gas prices on Ethereum are high.
Standardized Paymaster
Another new feature introduced by Grinta is a standardized paymaster interface. While Starknet had a paymaster in earlier versions, the interface was not standardized. The standardized interface brings compatibility across different dapps and crypto wallets, allowing devs to integrate with any paymaster without the need for custom logic which would vary on a per-vendor basis.
What’s Coming Next?
The remainder of 2025 will see Starknet continue on its journey of becoming a better-performing and more decentralized network. On the decentralization track, decentralized validation is planned. This means that stakers will participate in the consensus and will vote on blocks. At least two-thirds of the stake will be required to sign off on blocks in order for them to be finalized. Only sequencers operated by StarkWare will propose blocks at that stage, with full decentralization set for a future version.
As for performance, Starknet will move beyond 1000 TPS with the introduction of Cairo Native and will significantly reduce proving costs through improvements to S-two. Additional features, including faster L1 finality, better block times, and more are currently under discussion with the community.
The name Grinta – meaning grit, resolve, or determination – captures our resolve to make Starknet the manifestation of the blockchain vision: a high-performance and widely-adopted economic hub that doesn’t compromise on security or decentralization.
