Sigma Prime published Lighthouse v8.3.0 at 10:12 UTC on 19 Oct 2026. The tag, named Simple Rick, is a consensus-client binary that already includes Hoodi’s Gloas fork epoch. Hoodi operators who run Lighthouse have a hard clock: 26 Oct 2026 at 17:42:48 UTC, epoch 132352, slot 4235264.
The same notes say the release is suitable for Ethereum mainnet and that Sigma Prime wants mainnet Lighthouse users to upgrade. That sentence is about today’s software, not about a live mainnet hard fork. Gloas is still unscheduled on mainnet. The project says another release, v9.0.0, will be required for that fork.
Ether was $2,425.21 on our CoinGecko snapshot at , up 0.32% over 24 hours. Bitcoin was $82,394, up 0.45%. Coinbase ETH-USD last traded at $2,419 around 11:10 UTC the same day (session high $2,443.86, low $2,325.48). This is an operator story, not a range break.
What happened
Lighthouse is a Rust Ethereum consensus client. v8.3.0 is a stable tag, not a release candidate. The notes list three network clocks for Gloas, the consensus-layer half of Glamsterdam:
- Sepolia: epoch 353024, 6 Oct 2026 at 13:53:36 UTC, marked activated.
- Hoodi: epoch 132352, 26 Oct 2026 at 17:42:48 UTC.
- Mainnet: not yet scheduled.
Hoodi operators must upgrade the beacon node and the validator client together, plus a compatible execution client, before that Hoodi timestamp. The Ethereum Foundation already posted those client pairings in the 16 Oct 2026 Glamsterdam Hoodi announcement. We covered that post as a testnet clock, not a mainnet go-live, in our 17 Oct Hoodi read. Today’s tag is the missing Lighthouse binary for that same clock.
The notes also flag undisclosed security fixes. Sigma Prime says it will publish details later. Until that write-up exists, treat the mainnet upgrade advice as a patch instruction, not as a named CVE story.
Other shipped mechanics in the same tag: the Fast Confirmation Rule (FCR), partial data columns enabled by default on all networks including mainnet, a database migration to schema v32, builder and validator changes for enshrined proposer-builder separation (ePBS) after Gloas, and a Linux pre-built binary that now needs glibc 2.34 or later.
DefiLlama’s Ethereum fee overview printed about $8.42 million in 24-hour fees when we pulled api.llama.fi/overview/fees/ethereum at 11:10 UTC on 19 Oct 2026 (about 2% below the prior 24 hours). Chain TVL on the same host’s /v2/chains print was about $50.0 billion. Those figures describe a live fee market. They do not by themselves tell you whether to run FCR.
Context
Glamsterdam splits in two. Amsterdam is the execution-layer name. Gloas is the consensus-layer name. The headline feature on the consensus side is ePBS: builder and proposer jobs that used to sit in sidecars move closer to the protocol. Lighthouse’s notes say the client now supports the related block-production flow, including payload timeliness committee duties, optimistic sync, and historical backfill.
Hoodi validator operators get extra homework that mainnet operators do not, yet. External builders move into builder_definitions.yml from the Gloas fork onward. Gas limits on Hoodi follow a network schedule unless an operator overrides them. The scheduled Hoodi default becomes 200 million gas at Gloas. Mainnet has no such schedule in this release; Lighthouse keeps a 60 million default there. Per-validator settings and an explicit --gas-limit still win. If you need a refresher on why gas limits show up in user costs, see our gas fees glossary and the Ethereum history page.
FCR is the other new knob. The Lighthouse Book page for FCR and the v8.3.0 notes describe the same idea: under normal network conditions the rule can confirm blocks in about 1–2 slots (12–24 seconds). Without it, confirmation waits about two epochs, or roughly 13 minutes. Ethereum proof of stake finality does not change. When FCR is on, Lighthouse feeds the FCR-confirmed execution block as the execution layer’s safe_block_hash. Apps that trust the safe tag therefore see a faster “safe” head. Apps that need the strongest guarantee should keep using finalized.
Sigma Prime ships FCR off by default. The enable flag is --enable-fast-confirmation. The notes currently recommend it only on non-validating beacon nodes (RPC, wallets, bridges, indexers). Leave it off on beacon nodes that serve validators while the feature gathers mainnet hours.
Upgrade friction is real. Schema v32 migrates automatically on startup. Downgrade to v8.2.x is supported only before Gloas, using the v8.3.0 binary to walk the database back to schema v29, and not on Sepolia. Pre-built Linux binaries drop Ubuntu 20.04, Debian 11, and other glibc-older hosts; Docker is the workaround. Source builds now want Rust 1.91.0 and clang. Builder API users lose the old submitBlindedBlock v1 fallback and must speak V2. Web3Signer users need 26.9.0 or newer before the Hoodi fork.
This is a different object from Lighthouse v8.2.2 in August, which was a high-priority beacon-node patch. v8.3.0 is the Hoodi Gloas train, plus FCR, plus unnamed security fixes. It is also different from client tags that only advertised the same Hoodi epoch without a Lighthouse stable.
Our read
Run the binary. Do not trade the fork.
I care about shipped mechanics. v8.3.0 ships three things that matter on different calendars. First, Hoodi Lighthouse operators now have the consensus binary that matches the 26 Oct 17:42:48 UTC clock. Missing that upgrade is an operator failure, not a market event. Second, mainnet Lighthouse users are told to upgrade now for security and for schema/data-column defaults, while Gloas itself remains a later v9.0.0 problem. Third, FCR is real code with a conservative default. That is the right posture: show the faster safe tag to read-only infrastructure first.
I am not treating this tag as a mainnet Glamsterdam date, and I am not updating our 17 Oct Hoodi stance. The Foundation still has not scheduled mainnet. Sepolia already forked. Hoodi is the public rehearsal. Ether’s quiet 24-hour print around $2,425 does not need a Glamsterdam narrative glued onto it. For how we separate a live tape from a protocol calendar, see how to read a crypto market snapshot and the rest of Ethereum news.
Falsifiable claim: By 26 Oct 2026 23:59 UTC, Sigma Prime still presents v8.3.0 (or a later 8.3.x patch) as the Hoodi Gloas-ready Lighthouse release with activation at 26 Oct 2026 17:42:48 UTC, and still describes v8.3.0 as suitable for mainnet while stating that mainnet Gloas needs a later release. If the v8.3.0 notes retract the Hoodi epoch, pull the mainnet-upgrade recommendation, or relabel this tag as not Hoodi-ready, this read is wrong.
What to watch next
Watch the remaining blank execution-client seats for Hoodi as the 26 Oct 17:42:48 UTC mark gets close. Geth’s latest stable tag at scan time was still v1.17.8 (8 Oct 2026). A mainnet Glamsterdam blog post from the Foundation, or a Lighthouse v9.0.0 tag that actually sets a mainnet Gloas epoch, is the rewrite trigger. Client chatter on All Core Devs Testing is not.
Watch Sigma Prime’s promised security write-up. Until those issues are named, “upgrade mainnet” is an instruction with an unknown severity. If the disclosure says the bugs were not remotely triggerable, the urgency drops. If it says they were, the 19 Oct tag was the patch window.
Watch FCR in production the slow way: whether RPC providers actually flip --enable-fast-confirmation, and whether anyone reports safe tag surprises. The feature should stay off on validating nodes until Sigma Prime says otherwise.
Last, watch the Hoodi gas-limit jump to 200 million only on Hoodi. If someone quotes that number as a mainnet default from this release, they misread the notes.