Parachain finality is moving to a JAM service. The Relay Chain is being reduced to a coordinator. The projects are further down.
Gray Paper0.8.0
Published 3 June 2026. Parity named v1.0 as the audit target. No tag yet.
PolkaJAM0.1.28
Released 1 June 2026. Production target is a full node against Gray Paper 1.0, not the mainnet binary.
SwitchoverUndated
The 10 September note says finality is approaching, and guides come before any required upgrade.
Active validator
Sign the account you run in the set. The node stays as it is.
Polkadot.js, Talisman, SubWallet
Five pieces of work
Posted by @joyce for Parity Technologies. These are the deliverables. The post does not mark any of them shipped.
01
PolkaJAM Production client
The target is a full node on Gray Paper v1.0, built for a production network. It has to handle malicious peers and denial-of-service, stay orderly when overloaded, and restart safely at any time.
The same project includes an in-browser light-node library, so applications can read JAM directly and check the result.
02
Minimal Relay AHM v2
The Relay Chain becomes a minimal, transactionless coordinator. This finishes the Plaza Asset Hub Migration from the year before.
Parachain registration, channel management, and on-demand coretime ordering move to the Coretime chain. Users deal with them there. A migration guide is planned, with a map of each affected call and replacement tooling.
Deposits move with that functionality to the Coretime chain. Vestigial free balances move to Asset Hub.
The end state in the note: zero DOT on the Relay Chain, no user-callable transactions, and no way to move funds onto it again.
03
Offchain messaging Not JAM state
The note specifies a new messaging protocol for JAM. Messages stay off-chain and are not written into JAM state. It aims to match HRMP latency in the Polkadot SDK, then beat it.
On Polkadot, offchain messaging is added beside on-chain HRMP. Later work targets hundreds of milliseconds. The note calls that a target. It is not a mainnet measurement.
04
Parachain Service Finality on JAM
The target is audited, hardened service code on JAM. Parachains produce blocks and pass messages there. A switchover was prototyped and shown on a testnet. Mainnet has not switched.
The service covers what the Relay Chain parachain host does today. Block validation, code upgrades, key handling, coretime, and authorization.
05
The specification Alongside the client
Related work, listed separately. Finalise and audit Gray Paper v1.0. Then launch an initial JAM production network with nodes run by independent operators.
The public PDF is still 0.8.0. The note names a 1.0 audit. It names no firm, no report, and no date.
What moves, what does not
JAM does not take staking. @bkchr, 23 September: incentives stay in the upper layer, the logic called Polkadot today. JAM "will broadly work as the relay chain today." Services can still be upgraded without a fork. Raising the core count still needs a node change, same as now.
Taken from the 10 September note and the replies through 23 September. "Unchanged" means unchanged by this note.
Subject
After this plan
Election, rewards, commission
Upper Polkadot layer, not JAM. Unchanged. @sigurpol, 23 September, said election design, metanodes, and geographic spread are a separate discussion. Not specified yet. Not part of JAM.
Service business logic
Still forkless to upgrade. @bkchr: JAM does not remove that. Core count and similar changes still need a node release, as they do on the relay today.
Parachain validation, upgrades, keys, coretime
Parachain Service on JAM. This is the finality switchover the note says is approaching.
Registration, channels, on-demand coretime
Coretime chain. Wait for the migration guide before retargeting calls.
Relay Chain balances
Deposits follow the moved functions to the Coretime chain. Leftover free balances go to Asset Hub. End state is an empty relay: no DOT, no user transactions, no deposits back.
Cross-chain messages
Offchain messaging is added beside HRMP. JAM itself uses that path. The note does not remove HRMP.
The binary you run today
Stays until a guide says otherwise. PolkaJAM 0.1.28 is a client release. It is not a mainnet rollout.
Hardware and geography are outside the deliverable list. @burdges, 21 September: validators and cores still follow ELVES, with small changes. A rough reading is about three validators per core. Practice is short of that arithmetic. QUIC in JAM should help over later iterations, not in the first cut. A much heavier machine spec is possible and is not the goal. The set is meant to include individuals and weaker links, not only well-peered institutions.
Regional latency was raised on the thread. Missed votes, fewer era points, a European cluster. @bkchr called it a defect of the current protocol and of incentives. JAM is not specified to fix it. This is not a reward reform and not a new hardware minimum.
Before a guide exists
Parity said some of this may need an upgrade. They will say so. Support channels and guides go on the same thread over the coming months.
Subscribe to The Road to JAM. Parity named that thread for guides and support.
Keep the current mainnet client. The PolkaJAM 0.1.28 archives are not a replacement for it.
If your tooling sends relay extrinsics for registration, channels, or on-demand coretime, those calls move to the Coretime chain. Wait for the migration guide before you retarget them.
Read Gray Paper 0.8.0 for the spec that is published. v1.0 is the audit target named on 10 September.
Sources
The 10 September note, and the replies through 23 September.
NoteThe Road to JAM, @joyce, Parity Technologies, 10 September 2026
Replies@burdges, 21 September; @bkchr and @sigurpol, 23 September, on the same thread