Karst readies the OP Stack for Osaka

Optimism has documented Upgrade 19, named Karst, as the next proposed hard fork for OP Stack chains. The official schedule says Optimism-governed Sepolia chains will activate on June 17, 2026 at 16:00:01 UTC, while mainnet OP Stack chains are planned to activate on July 8, 2026 at 16:00:01 UTC, pending Optimism Governance approval. The named chains include OP, Soneium, Ink, Unichain, Mode, Metal and Zora. For operators, the immediate point is simple: clients need to be updated before the hard fork timestamp.

Karst matters because it is not only a date on the release calendar. The upgrade introduces the L2 Contract Manager, or L2CM, a mechanism for upgrading L2 predeploy contracts through a more structured and auditable consensus-layer path. In the OP Stack architecture, predeploys are system contracts that provide core functions, such as messaging between L1 and L2. Changing them cleanly matters for future protocol work, including interop between chains, because a poorly coordinated system-contract upgrade can affect withdrawals, bridges and applications that call those contracts directly.

The upgrade also brings part of Osaka to L2. Optimism’s notice lists EIP-7825, which sets a per-transaction gas limit of 16,777,216 gas, along with gas-cost changes for MODEXP and P256VERIFY, two cryptographic precompiles used by some smart contracts. A precompile is a low-level protocol function that executes an expensive operation more efficiently than ordinary contract code. In practice, applications that submit very large L2 transactions or rely on hardcoded gas assumptions for these functions should check their behavior before activation.

The most sensitive change is in fault proofs, the mechanism used to challenge an incorrect L2 state. Karst makes CANNON_KONA the respected game type for withdrawals, putting the Rust-based kona-client on the primary proof path. Optimism also states that op-geth and op-program will not support Karst, so operators need to migrate to op-reth and kona-client. The useful takeaway is operational rather than promotional: the OP Stack is moving toward a more modular protocol stack, but that modularity raises the cost of lagging behind. When one hard fork combines gas rules, system contracts and withdrawal proofs, node operators and application teams need to treat release tracking as part of production risk management.