

Sei v6.7 passed governance on October 8, 2026, preparing the FlatKV storage migration and advancing Sei’s roadmap toward Giga.
Author: Akshay
High attention and emotional sentiment detected.
8th October 2026 – Sei validators approved the Sei v6.7 upgrade on October 8. They passed governance proposal #125 on mainnet with zero No and zero NoWithVeto votes. The chain halts and upgrades at block height 236,450,000, expected near 16:30 UTC the same day.
High Signal Summary For A Quick Glance
Zerhylean
@Zerhylean
@SeiNetwork this is the step that unlocks the FlatKV storage migration. One cleaner path toward Giga’s higher throughput and lower latency.
The proposal to upgrade Sei to v6.7 is live. It prepares Sei for migration to a much faster EVM storage layer. One step closer to Giga, the ultimate trading blockchain. https://t.co/ojzMhD0qLE
12:20 PM·Oct 8, 2026
Blue Guy
@talhaaeth
@SeiNetwork storage layer upgrades rarely get this much hype. Curious what the actual throughput gain looks like post migration.
The proposal to upgrade Sei to v6.7 is live. It prepares Sei for migration to a much faster EVM storage layer. One step closer to Giga, the ultimate trading blockchain. https://t.co/ojzMhD0qLE
12:17 PM·Oct 8, 2026
The vote clears the way for one of Sei’s biggest technical changes yet. v6.7 does not flip the new storage engine on by itself. Instead, it loads the tooling and a governance trigger. A later vote can then move Sei’s EVM state to a much faster layer.
The Sei v6.7 upgrade is a mandatory binary release, tagged v6.7.0, that every validator must run. Sei Labs also shipped a recommended follow-on, v6.7.1, ahead of the storage work.
First, v6.7 continues SIP-3, Sei’s cleanup of legacy Cosmos code. It removes IBC core, IBC transfer, the capability module, and interchain accounts, along with related CosmWasm execution paths. The chain still decodes old transactions, so history stays readable.
Second, and more importantly, v6.7 adds the plumbing for a storage migration. That includes FlatKV commit and snapshot tooling, a lattice hash called LtHash, and a new block and receipt store. It also adds a governance-based trigger, so operators can activate the switch later through another vote.
According to Sei Labs, the release “sets up the future migration of EVM state from memIAVL to FlatKV.” The key word is future. The live commit store does not change today.
Voting on proposal #125 opened on October 5 at 10:31 UTC and ran for the standard three days. It closed on October 8 at 10:31 UTC, and the proposal reached PROPOSAL_STATUS_PASSED.
The on-chain tally shows roughly 2.53 billion SEI voting Yes, according to Seistream’s API. No and NoWithVeto both sat at zero, while abstain came to just 2.16 SEI.
Sei’s mainnet governance rules set a 33.4% quorum and a greater-than-50% approval threshold, excluding abstain. A 33.4% NoWithVeto share can also kill a proposal. With no opposition recorded, none of those brakes came into play.
The proposer staked the minimum 3,500 SEI deposit. Notably, an earlier proposal #124, also labeled v6.7, was rejected before this one passed. You can review the full record on the Seistream proposal page.
Every EVM node constantly reads and writes account balances, contract code, and storage slots. It then proves that the resulting state matches what validators agreed on. That proof step is where the current design strains.
Today Sei commits EVM state through memIAVL, a Merkle tree. Changing one value forces the node to recompute hashes all the way up to the root. As state grows, so does that cost, which becomes a wall at very high transaction volumes.
FlatKV takes a different approach. It stores state as a flat key-value map, so one logical change becomes one write. A lattice hash then keeps a running fingerprint that updates in constant time. Verification no longer scales with total state size.
In an internal Sei benchmark reported by BSCN, FlatKV averaged about 205,913 transactions per second over 24 hours. That compares with roughly 15,000 for memIAVL. The figure measures the storage engine alone, not end-to-end chain throughput. Treat it as a lab result, not a live promise.
Giga is Sei Labs’ multi-phase rebuild of the chain’s consensus, execution, and storage layers. Co-founders Jeff Feng and Jayendra Jog pitch it as a multi-proposer EVM Layer 1 built for trading.
The storage piece started landing in v6.6 on August 4, at height 224,201,091. That release made the Ares execution engine the default and began splitting EVM history into its own database. It also cut pruning time from 8 to 18 minutes down to about 5.
The Sei v6.7 upgrade is the next rung. Sei Labs marketing lead Campbell framed it bluntly on X. He called it a “hugely technical, extremely important piece of the Giga upgrade.” Jog added, “Another major upgrade on the road to Giga. Ship fast and iterate quickly.”
Still, independent reviewers urge caution on the broader roadmap. Research firm Common Prefix has flagged stake concentration and bad-case consensus costs. And Sei’s headline figures of 200,000 TPS and 5 gigagas per second remain design targets, not sustained mainnet results.
Sei’s path to Giga is a sequence of in-place upgrades on pacific-1, progressing from the v2 parallelized EVM to Giga components, storage migration and an eventual full Giga mainnet.
Sei launches its Cosmos-SDK trading-focused mainnet on pacific-1. The network does not yet include EVM support.
Sei opens its v2 public devnet to test the parallelized EVM before the mainnet upgrade.
Governance activates EVM compatibility, optimistic parallelization and SeiDB on the existing pacific-1 chain.
Sei v2 mainnet beta opens to everyone, making Sei’s parallelized EVM broadly available after its infrastructure-focused alpha.
Sei Labs publishes the Giga whitepaper, outlining a multi-proposer architecture targeting 5 gigagas, roughly 200,000 TPS and sub-400 ms finality.
The updated Giga design reports internal testing on a 40-node, 20-region devnet with sub-250 ms ordering finality and higher throughput targets.
At block 224,201,091, v6.6 makes Ares the default execution path and begins moving EVM history into Eidos, marking the first major Giga components on mainnet.
FlatKV storage testing reports roughly 205,913 TPS at the engine level, while a full Giga architecture test reaches about 63,200 requests per second and 1.33 gigagas per second. These are benchmark results, not mainnet throughput.
Governance proposal #125 opens voting to prepare Sei for the future migration of EVM state from memIAVL to the faster FlatKV storage layer.
Proposal #125 passes with approximately 2.53 billion SEI voting Yes and no No or No-with-Veto votes.
The mandatory v6.7 binary upgrade executes at block 236,450,000, preparing the network for the subsequent memIAVL-to-FlatKV storage migration.
A separate governance proposal is expected to initiate the memIAVL-to-FlatKV migration. The timing remains conditional on that proposal passing.
Full Giga mainnet remains undated and is expected to progress through phased milestones including Autobahn, Hermes and Sedna. The headline performance targets remain roadmap and internal-test targets.
After the halt at block 236,450,000, validators run the new binary and the chain restarts. Downtime should last only a few minutes. Exchanges including Bybit flagged a brief pause on SEI deposits and withdrawals around the switch.
The actual storage cutover is a separate step. Sei’s v6.7.1 notes point to a follow-up migration proposal, expected around October 9 near 17:00 UTC. That vote would finally move EVM state onto FlatKV through the new governance trigger.
SEI traded near $0.070 around the upgrade, up a modest 1.5% to 2% on the day. No sharp spike followed the news. Market watchers can track price on CoinGecko and chain TVL, near $33.5 million, on DefiLlama. None of this is financial advice.
For now, the Sei v6.7 upgrade is a preparation step, not the finish line. The real test arrives when the migration vote lands. Then Sei must prove its storage claims on a live validator set.
Our Crypto Talk is committed to unbiased, transparent, and true reporting to the best of our knowledge. This news article aims to provide accurate information in a timely manner. However, we advise the readers to verify facts independently and consult a professional before making any decisions based on the content since our sources could be wrong too. Check our Terms and conditions for more info.