Starknet v0.14.3 Explained: Dynamic Gas Fees, 30% Cost Cut, and What Builders Need to Know

Starknet v0.14.3 goes live on mainnet today, June 22, 2026, delivering one of the most significant fee-model overhauls in the Ethereum Layer 2 network's history. The upgrade introduces a dynamic L2 gas base fee mechanism, cuts target gas per block by 30 percent, and shortens block production times — all while deprecating RPC 0.8 in favor of newer API versions.

For builders deploying on Starknet or evaluating Layer 2 options for their next project, this upgrade reshapes the cost structure, developer experience, and performance profile of one of Ethereum's leading validity rollups. Here is everything you need to know.

What Is Starknet v0.14.3?

Starknet is a validity rollup (also known as a ZK-rollup) that settles on Ethereum. It uses STARK proofs to verify off-chain computation, enabling high throughput and low transaction costs while inheriting Ethereum's security guarantees. The v0.14.3 release is the latest in the 0.14.x series and focuses on making the network's fee mechanism more responsive to real-time demand.

The upgrade was tested on Starknet's Sepolia testnet starting June 9 and activates on mainnet today. It does not require any action from end users, but developers running infrastructure, full nodes, or applications that interact directly with the Starknet RPC will need to prepare.

Dynamic L2 Gas Base Fee: How It Works

The headline feature of v0.14.3 is the introduction of a dynamic L2 gas base fee adjustment mechanism. Previously, the minimum base fee on Starknet was set statically by the sequencer. This meant fees did not automatically respond to changes in network demand — they could be too high during quiet periods or too low during congestion.

With v0.14.3, the base fee adjusts automatically between blocks based on how full the previous block was relative to a target gas utilization. The design mirrors the core idea behind Ethereum's EIP-1559: when blocks are fuller than the target, the base fee increases; when they are below target, it decreases. This creates a self-regulating fee market denominated in STRK, Starknet's native token.

Since Starknet v0.14.0, all transactions have been required to specify STRK-denominated max_amount and max_price_per_unit values for each of the network's three fee categories. The v0.14.3 upgrade builds on this by making the floor price itself dynamic, so users benefit from lower fees during low-demand windows and the network remains protected against spam during high-demand periods.

30 Percent Reduction in Target L2 Gas Per Block

Alongside the dynamic fee mechanism, v0.14.3 cuts the target L2 gas per block by 30 percent. This does not reduce the maximum block size — blocks can still be as large as before — but it lowers the utilization threshold that the dynamic pricing algorithm uses as its equilibrium point.

In practical terms, this means the network will begin increasing fees sooner as blocks approach the new, lower target. The goal is to create a tighter feedback loop between demand and pricing, preventing sudden fee spikes while maintaining enough headroom for burst traffic. For most applications, the result is more predictable and stable transaction costs.

Shorter Block Times

The v0.14.3 upgrade also reduces block production times, meaning transactions confirm faster on Starknet's L2. While the exact block time varies depending on sequencer load and transaction complexity, the upgrade tightens the timeout window that the block proposer operates within.

Faster blocks improve user experience across the board — from DeFi swaps and NFT mints to gaming interactions and payment flows. For builders, shorter block times also mean that state changes propagate more quickly, enabling more responsive dApp architectures.

RPC 0.8 Deprecation: What Developers Need to Do

Starting with v0.14.3, RPC version 0.8 is no longer supported. Full nodes and SDKs will not release new versions compatible with RPC 0.8. Applications and tooling that still use this API version must migrate to RPC 0.9 or, ideally, RPC 0.10.1, which is the recommended version.

This deprecation affects developers who interact directly with Starknet nodes through raw RPC calls, as well as those using older versions of popular SDKs like starknet.js, starknet-rs, or cairovm-based toolchains. The migration path is straightforward in most cases — the newer RPC versions are backward-compatible for common operations — but teams should review the pre-release notes on the Starknet community forum to identify any breaking changes specific to their integration.

Why This Matters for Web3 Builders

Starknet's v0.14.3 upgrade reflects a broader trend across Ethereum's Layer 2 ecosystem: the shift from static, sequencer-controlled fee models to dynamic, market-driven pricing. This matters because predictable fees are one of the top factors developers consider when choosing where to deploy.

A dynamic fee model means builders can design gas-aware UX flows with greater confidence. Wallet integrations can show more accurate fee estimates. Protocols that batch transactions — such as account abstraction bundlers or intent-based systems — can optimize their submission timing based on real-time fee signals rather than guessing at sequencer behavior.

Combined with the 30 percent gas target reduction and faster block times, Starknet is positioning itself as a more mature and cost-efficient execution environment. For teams building cross-chain applications or choosing between L2 deployment targets, these infrastructure improvements lower the operational cost and UX friction of building on Starknet.

If you are building across multiple chains and need infrastructure that scales with your deployment strategy, thirdweb offers developer plans at https://thirdweb.com/pricing that support multi-chain applications out of the box.

What Comes Next for Starknet

The v0.14.3 release is part of Starknet's broader roadmap toward greater decentralization and performance. The dynamic fee mechanism lays groundwork for future governance over fee parameters, and the ongoing RPC version progression signals StarkWare's intent to stabilize a long-term API surface for the ecosystem.

Meanwhile, Ethereum's own L1 scaling roadmap — including the upcoming Glamsterdam upgrade with its 200 million gas limit target — creates tailwinds for L2s like Starknet. As Ethereum expands its data availability capacity, rollups can post proofs and data more cheaply, compounding the cost savings that L2-native fee optimizations like v0.14.3 deliver.

The Bottom Line

Starknet v0.14.3 is a meaningful infrastructure upgrade that makes the network cheaper, faster, and more responsive to demand. The dynamic STRK gas fee mechanism brings EIP-1559-style pricing to Starknet's L2, the 30 percent gas target cut tightens the feedback loop between demand and cost, and shorter block times improve confirmation speed across the board.

For developers: migrate off RPC 0.8 if you have not already, and take advantage of the more predictable fee environment to optimize your transaction flows. For the broader ecosystem, this is another step toward the kind of performant, cost-efficient Layer 2 infrastructure that mainstream web3 adoption demands.