Filecoin’s Solstice upgrade activates today: Fil+ ends and block rewards start funding paid storage

Filecoin is set to switch off its Filecoin Plus data-subsidy programme today. The network’s Solstice upgrade (network version 29) is scheduled to activate at epoch 6,470,279, which falls at 12:59:30 UTC on 19 October 2026, according to the Lotus v1.37.0 release notes. Its single protocol change, FIP-0118, ends the DataCap system and divides every block reward between the storage providers who win blocks and a new blockchain that runs exactly as written when it is called, holding and moving funds according to its code, without an intermediary able to change the outcome.">smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">pool for “Service Orchestrators”, with a burned remainder that only opens in 2027. At launch, 95% of each reward still goes to block winners.

This Filecoin upgrade is a hard fork: nodes that have not installed a compatible client by the activation epoch will drop off the main chain. The Filecoin status page lists a maintenance window from 12:59 to 13:59 UTC. This article was written on the morning of 19 October, before activation: the switch-over is due at about 13:00 UTC today, and nothing here describes how it went.

It lands four days after another supply event. The six-year vesting of the 400 million FIL set aside for Protocol Labs and the Filecoin Foundation ended around 15 October. Crypto Briefing, using its own arithmetic, estimates that gross new FIL issuance falls by roughly 75% as a result; that figure belongs to the vesting story, not to FIP-0118. FIL traded near $1.05 at 09:19 UTC, down 0.8% over 24 hours, with a market value of about $876 million, according to CoinGecko.

Solstice activates at 12:59:30 UTC: the timetable

Filecoin produces a block epoch every 30 seconds, so upgrades are pinned to an epoch number rather than a clock time. The Lotus team gives epoch 6,470,279 as 12:59:30 UTC, which is why the switch is usually rounded to 13:00 UTC. The same epoch and time appear in the upgrade metadata merged into the Filecoin core-devs repository on 6 October, along with built-in actors version 19.0.1.

Nodes start a pre-migration 120 epochs ahead. At 30 seconds an epoch that is one hour, so the first work begins at about 11:59:30 UTC. On the Lotus team’s benchmark machine (64 threads, 503 GiB of RAM, NVMe storage, measured on 18 September) the pre-migration took about 64 seconds and the migration at the upgrade epoch about 63 seconds, with daemon memory peaking near 38 GiB. The release notes warn that slower disks or busier machines will take longer.

The code has been running on Filecoin’s Calibration test network since 28 September, when it activated at epoch 4,109,133 (12:59:30 UTC) using Lotus v1.37.0-rc1, released on 23 September.

FIP-0118 ends Fil+ and DataCap without a wind-down

Filecoin Plus, usually written Fil+, has been the network’s way of steering rewards toward storage of real client data since its early years. Clients approved by a chain of human reviewers (root key holders, then “allocators”) received DataCap, and sectors filled with that verified data earned ten times the quality-adjusted power (QAP) of empty “committed capacity” sectors. More QAP means a bigger chance of winning blocks, so verified deals were worth far more to a storage provider.

The FIP-0118 text, created on 14 July and now marked Accepted, says that design could not work at scale because of “the impossibility of programmatically detecting ‘useful data'”. The authors, who include Molly Mackinlay and Rod Vagg, add that the review pipeline slowed onboarding and left the system open to fake client demand.

From activation, DataCap goes away in one step. No new DataCap is minted, unused balances become inert, pending allocations can no longer be claimed, and the verified registry and DataCap actors refuse any state change. The FIP chose an immediate cut over a phase-out because the old pathways would add complexity without changing anyone’s rewards.

In place of the 10x bonus for verified data, every new sector gets the 10x multiplier by default, whatever it holds. Existing sectors keep the power they have today; nothing is changed retroactively.

What changes for storage providers

Old committed-capacity sectors sit at 1x. Their owners can lift them to 10x in two ways: by snapping real data into them with SnapDeals, or with a new batch method, UpgradeSectorQuality, that raises power without new data. Either way the sector’s pledge collateral rises to the level of a 10x sector, so providers will need FIL on hand to lock. The method also lets a sector’s expiry be extended in the same call.

The FIP authors estimated the effect using data from 4 July. Committed-capacity sectors then held about 208 PiB of raw storage. If all of it moved to 10x, network QAP would rise by about 1,872 PiB, from roughly 14,760 PiB to 16,632 PiB, a one-time jump of about 12.7%. For scale, they note the network was losing 20 to 37 PiB of QAP a day, net, so the full bump equals two to three months of that decline.

Client software changes follow. Lotus retires its lotus filplus DataCap commands and adds lotus-miner sectors upgrade-quality. Curio v1.29.0, released on 12 October as the required Curio version for the upgrade, rejects new verified MK12 deals and allocation-based MK20 deals and offers sptool sectors upgrade-quality for use after activation. Its notes say that deal intake stops as soon as the new version runs, before the fork itself.

Block rewards split three ways: consensus, service, burn

The second half of FIP-0118 is the bigger economic change. Until now, almost the whole block reward went to the provider that won the block. From today, the reward actor divides it into streams with set weights. The consensus stream (called w1) keeps paying block winners. The service stream (w2) pays registered Service Orchestrators. The remainder (w0) goes to the burn address, where the FIL is destroyed. The Filecoin blog post that introduced the plan on 17 July put it this way:

“Whatever share leaves consensus but has not been earned by the service stream is burned.”

The consensus weight starts at 95% and declines in a straight line to a 50% floor over nine quarters. FIP-0118 defines a quarter as 262,974 epochs, or about 91.3 days, counted from the activation epoch. By our calculation (nine times 262,974 epochs of 30 seconds, added to today’s activation), the floor is reached around 18 January 2029.

The service share starts at 5%. During the first quarter, a bootstrap period, it rises in step with the consensus decline and reaches 10% at the start of the second quarter, with no condition attached. The FIP schedules no burn share during that first quarter. By the same calculation, quarter one ends around 18 January 2027, so that is the earliest date the burn stream can open.

From the second quarter on, the service share can rise by 5 percentage points per quarter, and only when Filecoin Pay, the network’s on-chain payment system for storage deals, settles enough volume through the Orchestrators. The first target, for the step from 10% to 15%, is $3,500 of payments in a quarter. The FIP says that figure was calibrated on current Filecoin Pay activity at a FIL price of $1. Each further step needs 2.7 times the previous target, up to about $3.66 million for the last step, from 45% to 50%.

Whatever sits between the two weights is burned. If every gate passes on time, the FIP’s own chart shows the burn swinging between about 0.94% and 5.93% of each reward, because the service share steps up in jumps while the consensus share slides smoothly. The FIP says a failed gate leaves the service share where it is while the consensus share keeps falling, and the gap is burned. Our arithmetic from those rules: if no gate ever passes, the service share stays at 10% and, once consensus reaches its 50% floor, the burn takes 40% of every block reward.

One Orchestrator to start, and multisigs in charge

In this first phase, the Orchestrator registry starts with a single address. Neither the FIP nor the July blog post names the entity behind it. Orchestrators are meant to turn their share of rewards into paying demand, for example by funding client acquisition or migrations, and what they do with the FIL happens off chain. Each Orchestrator’s share of the service stream is set by the Filecoin Pay volume it can show, posted quarterly and open to a seven-day verification window.

Two new contracts run the system. The Stream Weights Actor sets the split between streams; a discretionary change to its gate settings needs a published FIP, approval from both of its multisig accounts, which the FIP says are held by independent entities, and a seven-day hold (20,160 epochs) during which either multisig can cancel. The Service Rewards Actor keeps the Orchestrator registry. Adding or removing an Orchestrator needs both registry multisigs but no FIP. Opening registration to anyone is planned for a second phase, once a standard for on-chain attribution has been written; there is no date for it.

FIL supply: vesting ended on 15 October, and burns come next

Solstice is easy to confuse with the larger supply change that came just before it. When Filecoin launched in October 2020, 300 million FIL for Protocol Labs and 100 million FIL for the Filecoin Foundation began to unlock in a straight line over six years. That schedule ended around 15 October, removing about 66.7 million FIL a year from new supply. Crypto Briefing calculated on 10 October that gross issuance would fall from about 88 to 89 million FIL a year to about 21 to 22 million, depending on mining pace, a cut of roughly 75%. That estimate comes from the outlet’s own arithmetic, not from the Filecoin teams. A second Crypto Briefing piece, on 16 October, put the same figure on the “combined” vesting and Solstice changes without saying how much comes from each.

The 10 October calculation does not mention FIP-0118 at all, so the projected 75% drop is a vesting figure. Solstice does not change how much FIL is minted per block; the FIP itself describes issuance as roughly 20 million FIL a year in an indicative chart. What it changes is who receives that FIL, and how much of it is destroyed. During the first quarter, nothing goes to the burn. After that, the burn depends on whether paid storage grows fast enough to pass the gates.

Net supply also depends on things neither change controls, such as transaction fee burns and the FIL that providers lock as collateral when they add sectors, which would rise if many committed-capacity sectors are upgraded to 10x. Readers who follow other networks’ supply rules can compare this with how TRX supply kept growing despite TRON’s burns, and our glossary entry on token unlocks explains why vesting schedules move supply forecasts.

Who is affected: providers, clients, FIL holders

Storage providers face the most direct change. A provider that relied on verified deals for its 10x power loses that edge, because every new sector now gets it. Providers sitting on large amounts of committed capacity can upgrade it, at the cost of extra pledge. Over two years, the block-reward slice that goes to winning blocks shrinks from 95% to 50%, which lowers the reward per unit of power unless the network’s total power falls with it.

Clients lose DataCap as a bargaining chip. Many storage deals were cheap or free because providers valued the 10x multiplier more than the client’s payment. FIP-0118 bets that payments through Filecoin Pay, routed via Orchestrators, will replace that subsidy. Clients with pending DataCap allocations should check with their providers, as unclaimed allocations become unusable at activation.

FIL holders get a different supply story. Less FIL flows to providers who may sell it to cover costs, and a share can be burned. Whether that matters for the price depends on the gate figures and on demand, which no one can know in advance. Nothing here is a forecast or investment advice.

Custodians often warn clients around Filecoin upgrades: for the previous fork in May, Safeheron advised avoiding FIL deposits and withdrawals during the upgrade window and said deposits could be delayed. Holders who plan to move FIL today should read their platform’s notice first. For more on how projects such as Filecoin differ from bitcoin and ether, see our guide to altcoins and their risks.

What node operators must run before epoch 6,470,279

The Lotus team’s instruction is plain: all mainnet node and storage provider operators “must upgrade to v1.37.0 before epoch 6470279.” In practice the team now points to later versions:

• Lotus v1.37.0 (6 October) is the minimum version for the fork.
• Lotus v1.38.0 (12 October), node and miner, is the recommended release. Its notes flag that MinerCreateBlock now needs the sign permission, that eth_call and eth_estimateGas now honour positive gas limits, and that the minimum Go version is 1.26.8.
• Lotus v1.38.1 (13 October) is a node-only patch, recommended for indexers and RPC nodes that serve Ethereum log queries.
• Forest v0.38.0 (8 October), the ChainSafe client, is described in its release notes as a mandatory release for mainnet node operators. It also caps JSON-RPC batches at 100 requests by default.
• Curio v1.29.0 (12 October) is the required release for providers on Curio.

API users have a few breaking changes to handle. StateMinerInitialPledgeCollateral returns an error after the fork, and the Lotus notes say to use StateMinerInitialPledgeForSector with the full sector size as the verified size. PreCommitSectorBatch2 rejects any message with non-empty deal IDs, and the sealing pipeline refuses such sectors when they are queued. Curio warns that from the fork it passes an allocation ID of zero, even when finalising uploads for deals accepted earlier, so contract integrations have to handle that value.

Filecoin’s last upgrades: Golden Week and Fire Horse

Solstice is the third mainnet upgrade in about thirteen months. Network version 27, Golden Week, activated on 24 September 2025 at 23:00 UTC (epoch 5,348,280), and the Filecoin status page marked the maintenance complete an hour later. Network version 28, Fire Horse, followed on 27 May 2026 at 22:00 Singapore time (14:00 UTC), according to a Safeheron maintenance notice.

FIP-0118 describes the Fil+ system as an implicit split of the block reward, steered through DataCap. Solstice replaces it with an explicit split written into the reward actor, and ties part of it to payments measured on chain.

What to watch after 13:00 UTC

The first check is simple: whether the chain moves past epoch 6,470,279 with the new actors and without a long stall, and whether the status page closes its maintenance window on time.

Then come slower signals. The pace at which providers run upgrade-quality on committed-capacity sectors will show up in network QAP and in locked collateral. The name of the first Orchestrator, and the governance repository the FIP says will open at activation, should show who controls the service stream in practice.

The first real test of the model comes later. By our calculation from the FIP’s quarter length, the second quarter after activation ends around 20 April 2027. Its Filecoin Pay volume decides whether the service share goes from 10% to 15%, and the FIP’s timetable means any step would apply about 17 days after the quarter closes, so around 7 May 2027 at the soonest. A $3,500 first target is low. The targets after it rise fast.

Readers new to these terms can start with our tokenomics glossary entry. Epochs, sectors and quality-adjusted power are Filecoin-specific, and the Lotus releases page will carry any patch if the fork needs one.