Optimism just shipped Karst releases for 10 chains. The real upgrade is the proof system.

Karst is not one release. It is five.

Optimism has shipped mandatory software for Upgrade 19: op-reth/v2.3.2, op-node/v1.19.1, kona-node/v1.6.1, op-contracts/v7.0.0, and op-deployer/v0.7.0. The hardfork is scheduled for July 8, 2026, pending governance approval. According to the op-node release notes, 10 Superchain chains are in scope: op, ink, zora, mode, lisk, metal, unichain, soneium, ethernity, and tbn.

The obvious story is the op-geth shutdown. The better story is that Karst also moves Optimism's proving layer forward. op-contracts/v7.0.0 includes a new ZK dispute game, SP1Adapter, EIP-7825 gas cap enforcement in NUT bundle validation, and shared dispute game migration for interop.

At the contract level, that means Karst is not just runtime cleanup. It is a step toward a multi-prover Superchain, where security does not depend on one proving path forever. The tradeoff is coordination risk for a cleaner execution and proof stack.

Operators still need to treat the client change as hard mandatory. Per the op-reth release, op-geth has reached end of support and operators must run op-reth for Karst. The op-node release adds the other half: older op-reth versions will also fail after Karst because op-node now requires a new engine API method, getPayloadV5.

If you run Superchain infra, this is not a casual upgrade window. A mismatched op-node and op-reth pair can break you after activation. This is a version-coupling problem, not a vibes problem.

There is also one known footgun. The release notes flag it: eth_estimateGas calls have a known issue after Karst. A fix ships in op-reth/v2.3.3 before mainnet activation. That fix is now the small release that matters most.

Read the PR diff, not the thread. Karst completes the op-geth sunset, activates Osaka at the L2 level, and quietly adds ZK proof infrastructure. The code does what the announcement says, plus the proving work that will matter more over time.