zkSync just shipped upgrade plumbing. The skipped version makes the next big upgrade feel close.
Matter Labs did not announce a hard fork date, but the code reads like a team reducing failure paths before one.
CryptoVibe Desk · zksync · matter-labs · protocol-upgrades
- →Matter Labs published zkSync Era core-v31.0.0 on June 18, jumping from v29.20.0 without a public v30 release.
- →The important parts are Airbender failure handling and pre-upgrade data commitment config, not the routine bug fixes.
- →Watch whether Matter Labs names the next upgrade window before July ends, because this release looks like staging work.
- Airbender → Airbender is Matter Labs' in-house system for generating the proofs that zkSync uses to check transactions.
- prover → A prover is software that creates a cryptographic proof showing a batch of transactions was processed correctly.
- l2_da_commitment_scheme → A configuration value that tells zkSync how to record a compact promise about where transaction data can be checked. It can now be set before the upgrade that activates it.
- EraVM bytecode → EraVM bytecode is the low-level program format that zkSync contracts run on.
zkSync Era jumped from v29.20.0 to v31.0.0 on June 18. Matter Labs' release notes list 2 new features and 5 fixes. The missing v30 is not explained in the notes.
That gap matters less than what shipped. This reads like pre-upgrade hardening, not normal cleanup. The code-facing changes target failure handling, retry limits, and upgrade config. Those are the places teams touch when they expect the next protocol change to matter.
The first real change is in Airbender. Matter Labs added proof-failure submission handling and bounded proving retries, according to the release notes. In plain English: if proof generation fails, the system now has a clearer path for reporting it. It also stops retrying forever.
That sounds boring until you're operating the chain. Unbounded retries are not resilience. They are a queue with no backpressure. If you're a zkSync app team, your bag depends on the prover failing in a predictable way, not failing loudly after everyone notices.
The tradeoff is simple. Matter Labs gives up infinite optimism for bounded failure. That is usually the right call in prover infrastructure. A failed proof is bad. A failed proof that keeps consuming operator attention without a clean state is worse.
The second change is more interesting. The l2_da_commitment_scheme can now be configured before the protocol upgrade that activates it. That means operators can prepare the value before the state transition depends on it. At the contract level, this reduces the upgrade window where one wrong config can create a messy mismatch.
This is not proof that a fork is scheduled tomorrow. The release notes do not say that. But teams usually do not pre-configure dormant upgrade fields for fun. They do it because the activation path is close enough to test around.
The fixes point the same way. Matter Labs lists API stability, contract verification, state_keeper mempool sync, and verifier behavior among the 5 fixes. The verifier now rejects partial matches for metadata-less EraVM bytecode. That closes an ambiguity where tooling could treat a near match like a real match.
That matters for users and explorers. Contract verification is a trust surface, even when nobody calls it that. If a verifier accepts a partial match, a user can think they are reading the deployed code when they are not. The numbers don't add up when verification is fuzzy.
Read the PR diff, not the thread. This release is defensive plumbing. The headline feature is not a faster chain or a nicer developer story. It is a narrower failure surface before the next high-stakes upgrade.
Matter Labs' choice to bound Airbender retries before activating the DA scheme is the right failure-mode work, because unbounded proving loops turn upgrade windows into operator incidents.
Before July 31, 2026, watch whether Matter Labs publishes an upgrade plan that names the l2_da_commitment_scheme value and Airbender retry limits.
Primary links and supporting reads used by the desk for this story.
Forward this.
