Sui testnet swapped validator votes for timestamps to advance epochs. Mysten is quietly removing a coordination layer before mainnet.
PR #26740 replaces a consensus step with a clock on testnet. That's not just cleanup. It's a signal about where Mysten wants epoch timing to land.
CryptoVibe Desk · sui · protocol · testnet
- →Sui testnet-v1.73.0 shipped Protocol Version 125, replacing validator-vote epoch advancement with timestamp-based logic via PR #26740.
- →Removing validator coordination from epoch timing cuts Byzantine overhead on testnet and signals how Mysten wants L1 clock behavior to work before mainnet.
- →Watch for a Sui Improvement Proposal formalizing timestamp-based epoch advancement for mainnet before Q3 2026.
- epoch → A fixed time window in Sui after which the validator set can change and staking rewards are settled.
- validator votes → A round of agreement between multiple network nodes required before a shared state change is allowed to happen.
- Byzantine agreement → A consensus method that works correctly even if some nodes lie or fail. Removing it from epoch timing makes the process faster and simpler, but trades away that fault-tolerance guarantee.
Sui testnet-v1.73.0 ships Protocol Version 125, per Mysten's release notes. One protocol change matters more than the update list suggests.
PR #26740 replaces validator-vote epoch advancement with timestamp-based logic on testnet. Before this, testnet epochs ended when validators coordinated a vote. Now the clock decides. That's a meaningful shift in how the network's timing layer works, even if it's testnet-only.
The tradeoff is coordination for predictability. Validator-vote advancement gives you a consensus check: validators agree before the epoch closes. Timestamp-based logic removes that round of Byzantine agreement and makes test cycles reproducible.
On testnet, that's the right call. You're not protecting production staking. You're running infrastructure that needs to behave cleanly across releases.
The engineering analogy: switching from a consensus-gated flush to a scheduled cron job. Simpler, faster, easier to reason about. But it does mean testnet epoch boundaries now land differently than mainnet. If you're building tooling that assumes epoch endings track validator signals, update that assumption now.
The rest of v1.73.0 is developer-facing cleanup. GraphQL gains checkpoint lookup by digest and an asTransactionObject query path. The CLI adds a sui move format passthrough command and graceful Ctrl+C handling for sui start --with-faucet. The indexing framework gets retry safeguards that will quietly matter the next time a remote RPC drops mid-sync.
The release shipped first as devnet-v1.73.0 on May 25, then graduated to testnet on May 26. That sequencing is standard for Mysten. It also means the changes ran on devnet for a day before testnet validators saw them.
The open question is whether timestamp-based epoch advancement is Mysten's mainnet design target or a testnet-only simplification. The release notes don't say. That's what engineers should be watching.
If you're running a Sui validator or building indexer logic against testnet epoch boundaries, read PR #26740. The change is small. The tooling implications are not.
Mysten shipped timestamp-based epoch advancement with no mainnet spec attached. Validator operators are now reading testnet releases to infer design direction. That is a credibility problem for a team trying to attract institutional node runners.
Watch for a Sui Improvement Proposal formalizing timestamp-based epoch advancement for mainnet before Q3 2026. If it ships, this is the mainnet design target, not a testnet-only simplification.
Primary links and supporting reads used by the desk for this story.
Forward this.
