A bug in staking means depositing a proof of stake coin with a protocol that stakes it for you and issues a transferable token (such as Lido stETH or rETH) representing the deposit plus rewards, which you can use or sell while the coins stay staked.">Reth, the Ethereum execution client built by Paradigm, could make a node compute a wrong state root, and the networks that run it shipped fixes between 4 and 6 November 2026. Base, BNB Chain’s BSC-Reth and the Tempo blockchain released patched versions, and Paradigm’s chief technology officer told operators of Optimism’s op-reth to upgrade as well. Base says it checked its current state root and that “users were not affected”. No source reports a loss of funds, an exploit or an outage.
Early headlines called it an Ethereum bug. That is too broad. The flaw sat in one client, in the path BNB Chain’s release note calls the legacy state-root calculation. It was not in the Ethereum protocol, and it was not in Geth, the most widely used execution client. No source names Arbitrum, which runs its own Geth-based Nitro software, as affected, and Arbitrum has made no statement.
The knock-on effects were precautions. LayerZero paused some cross-chain routes, and the risk manager Gauntlet restricted deposits into some vaults, according to press reports. As of 07:30 UTC on 7 November, LayerZero had published no statement of its own on the pause or on when routes would restart.
The Reth bug: a node could get the state root wrong
Every block on Ethereum, and on the layer-2 networks built on it, commits to a state root: one hash that sums up every account balance and every contract’s storage once the block’s transactions have run. Each node recomputes it. A node that gets a different number disagrees with the chain. It may stop following it, or, if it builds blocks, produce one the other nodes reject. Our guide to what a blockchain node does explains the checks involved.
Base’s status page says the Reth bug “could, in some cases, make a node compute an incorrect state root”. BNB Chain’s release note places it in the legacy calculation path, “under specific trie conditions”. CryptoSlate went further on 6 November, writing that certain smart-contract storage layouts could set it off, that a crafted transaction could trigger it under specific conditions, and that block-building software used the vulnerable method on lower-end hardware and in OP Stack networks. No primary document confirms the crafted-transaction point. Whether anyone could have exploited the flaw on purpose is not known.
The worry went beyond one node falling out of step. Proof systems and some bridges rely on state roots to decide what happened on a chain, which is why Konstantopoulos’s warning named DeFi bridges and “state-proof backed systems”. He also said, according to CryptoSlate, that most bridges rely on receipt roots or multisig sign-off rather than the affected calculation. Base’s status page says its proof system checks every state root with an independent implementation. TokenPost called the flaw “a client correctness issue, not a known attack”.
A fix that did not say what it fixed
The code change came first, and quietly. Pull request #28264 on Reth’s GitHub, titled “fix(trie): tighten iterator seek boundary”, was opened at 20:13 UTC on 3 November and merged at 21:00 UTC. It changes one file, adding 79 lines and removing 12. Its description talks about how an iterator decides where to seek at a cached branch boundary, and it adds a regression test. There is no label and no word about security.
Base’s status page dates the discovery to 4 November, and Base says it applied a mitigation to all its block builders that day. On 5 November three releases landed within about 70 minutes. Paradigm described Reth v2.10.0 as a “hardening release” with high update priority for every kind of node, and listed the fix under Engine & Trie. The notes never use the words state root or security.
| Time (UTC) | Who | What |
|---|---|---|
| 3 Nov, 21:00 | Paradigm (Reth) | Pull request #28264 merged |
| 4 Nov | Base | Bug found, mitigation applied to its block builders |
| 5 Nov, 15:10 | Optimism | op-reth v2.8.1 released |
| 5 Nov, 16:01 | Base | Node v1.5.3: “Prevents a state root mismatch” |
| 5 Nov, 16:18 | Paradigm (Reth) | Reth v2.10.0, “hardening release” |
| 5 Nov, 17:52 | Base | Status page asks operators to update |
| 5 Nov, 22:17 | Georgios Konstantopoulos | Public warning on X |
| 6 Nov, 02:53 | BNB Chain | BSC-Reth v0.1.4-hotfix |
| 6 Nov, 18:13 | Tempo | v1.18.2, high priority for validators |
| 6 Nov, 18:53 | Paradigm (Reth) | v2.10.1 maintenance release, bug not mentioned |
Unchained counted about 43 hours between the merge and Base’s public notice. As of 7 November there was no GitHub security advisory on the Reth repository, no source cited a CVE number, and Paradigm had not published a post-mortem.
Konstantopoulos’s warning, and a release note that stays silent
The public alarm came late on 5 November. At 22:17 UTC Georgios Konstantopoulos, Paradigm’s chief technology officer, posted a PSA on X: a fix had gone out across the Reth ecosystem so that Reth-based nodes, DeFi bridges and state-proof backed systems “are all safe”. He asked anyone who depends on Reth to move to Reth v2.10.0, OP Reth v2.8.1 and Base v1.5.3.
CryptoSlate reported that in the same thread he described an edge case in Reth’s legacy state-root code, apologised for the disruption, said the withdrawal systems of Base and other OP Stack chains were not exposed, and said Optimism had received a patch. Coindaro could not read the full thread, so those points rest on CryptoSlate’s account.
One detail does not line up. The op-reth v2.8.1 notes tell all node operators to upgrade but describe only a fix to the eth_capabilities RPC call. They say nothing about state roots, although the release spans 153 commits since v2.8.0. So the claim that it carries the trie fix rests on Konstantopoulos’s post and on press reports. We found no Optimism status notice either; our page on OP Mainnet and the Superchain covers who maintains that software.
Base, BNB Chain and Tempo: who patched, and what they said
Base: no resync, and the halt risk removed
The Base v1.5.3 node release says it “Prevents a state root mismatch” and marks the update as recommended, with no deadline and no breaking change. Base’s status page opened a maintenance entry at 16:19 UTC on 5 November and, at 17:52 UTC, asked operators to update their nodes as soon as possible. It says no resync is needed and that the release removes the risk of Base validators halting on an invalid state root.
“We’ve verified the current state root is correct and users were not affected,” Base wrote. The entry lists a window running to 16:19 UTC on 7 November and shows no closing update. A separate 2 November delay in transaction inclusion, and the Denim upgrade of 5 November, are not tied to this bug.
BNB Chain: a hotfix for every BSC-Reth operator
BNB Chain published BSC-Reth v0.1.4-hotfix at 02:53 UTC on 6 November. The notes say it brings in the upstream Reth fix and needs no database migration or hard fork. They ask validators, MEV builders, full nodes and sentries running reth-bsc to upgrade. In a post on X quoted by crypto.news, BNB Chain said “opBNB does not use op-reth, and the impact on BSC-Reth is limited.”
Tempo: the team that may have found it
Tempo, the stablecoin payments blockchain incubated by Stripe, shipped v1.18.2 at 18:13 UTC on 6 November, with high priority for validators and a line saying it updates Reth with a fix for trie iterator seeks. Tempo is a layer-1 chain, not a layer-2; our explainer on the difference between layer 1 and layer 2 sets out why that matters.
CryptoSlate reports that the Tempo team found the bug through fuzzing, a test method that throws huge numbers of random inputs at software to make it fail. No other source we read says so, and Base’s own page records the discovery on 4 November without naming who made it.
LayerZero paused routes, and the bridges got nervous
LayerZero, the messaging protocol that many token bridges use, paused “certain cross-chain routes as a precaution” that “should resume soon”, according to a Frax post on X that Unchained quoted on 5 November. Frax has since deleted it. TokenPost wrote on 7 November that LayerZero paused parts of its off-chain infrastructure for some EVM networks and that messages were delayed, not lost. No LayerZero statement gives a start time, a list of routes or a restart. CryptoSlate notes that the pause does not show LayerZero’s contracts were vulnerable.
Gauntlet, the risk curator, restricted deposits into its V1 vaults, paused allocations and cut its exposure to LayerZero-bridged assets, according to The Defiant, whose summary says no exploit had been reported to Gauntlet’s knowledge. Unchained, relaying that report, added that withdrawals stayed open. Gauntlet’s own post on X has been deleted, so those details rest on the two outlets.
Bridges have reason to be jumpy. In April, the $292 million exploit of Kelp DAO’s rsETH bridge, which LayerZero tied to North Korea’s Lazarus Group, put cross-chain plumbing under suspicion. Unchained drew the parallel, and no source links the two events. Our LayerZero project page covers what the Kelp hack changed. ApeChain’s plan to move its assets from LayerZero to Chainlink CCIP was announced on 4 November, before the Reth warnings, crypto.news reported.
The second Reth state-root bug in 14 months
Reth has been here before. On 2 September 2025 a different state-root bug stalled some Ethereum mainnet nodes running Reth 1.6.0 and 1.4.8, The Block reported the next day. Konstantopoulos shared recovery commands then too. The Block cited Ethernodes data putting Reth on 5.4% of execution-layer clients at the time, and said only a small share of nodes seemed to be hit. No source gives a current figure.
A small share is what lets Ethereum mainnet absorb a bug in one client: nodes on Geth and the other clients keep computing the right root and the chain moves on. A layer 2 whose block builders all run the same client has no such cushion, which is why Base pointed to its independent proof implementation when it said users were not affected.
Proof software has had its own scares this autumn. In October, Succinct fixed an SP1 bug that let invalid proofs pass its Rust verifier, with on-chain checks unaffected.
What is still unknown, and what to watch
Paradigm has published no post-mortem, no advisory and no technical description of the exact trigger. Nobody has said whether the bug was ever hit on a live network, how many nodes ran affected versions, or whether OP Stack chains such as Unichain or Ink were hit or only patched.
The next markers are a post-mortem or security advisory from Paradigm and a word from LayerZero that its routes are running again. Base’s maintenance window closes at 16:19 UTC today.
Nodes left on old code are the remaining risk. Blockaid warned, as relayed by Unchained, that they could freeze or return faulty state proofs. The projects have asked operators to move to Reth v2.10.0 or later, op-reth v2.8.1, Base v1.5.3, BSC-Reth v0.1.4-hotfix or Tempo v1.18.2.