Daily digest of the Bittensor ecosystem

中文/EN

Daily edition
Oct 23, 2026

Bittensor Daily A 1T1B product
1T1B ↗
BITTENSOR DAILY
Network upgrades·eTao·

24H Market

01/10

Gainers
--Loading…--
Losers
--Loading…--

THE DAILY BRIEFING

Bittensor Daily — October 23, 2026

SN23 (Trishool) introduced Astroware, a separate commercial entity that buys continuous red-teaming work from the subnet and is meant to fund alpha buybacks once revenue begins flowing. Two breaking protocol changes tied to a runtime spec update landed in the codebase, covering root claims eligibility and stake reporting during subnet dissolution. Elsewhere, SN92 (REAX) resolved a validator emissions bug and reversed its playoff ruling to name three co-champions, while SN6 (Numinous) launched a new graph-based forecasting competition.

Today's highlights

  1. 01

    Pull requests 3276 and 3101 introduce breaking changes tied to a runtime spec version update, covering root claims for baskets holding all 128 subnets plus TAO and continuous stake reporting through subnet dissolution.

    Node operators and validators need to run the updated version, and anyone consuming stake reporting data during dissolution is affected.

  2. 02

    SN23 (Trishool) launched Astroware, a separate commercial entity under the same founding team that buys continuous red-teaming work from the subnet and introduced a product called HaloGuard EU, with the team saying the subnet will receive a share of Astroware's revenue to initially fund alpha buybacks once they start.

    It establishes a direct commercial revenue link between an off-chain product and on-chain token buybacks, a model other subnets may watch.

  3. 03

    SN92 (REAX) fixed a validator bug that had zeroed emissions for several engine-track UIDs, reversed its initial single-champion decision to name uid 250, uid 116 and uid 199 as co-champions after finding a statistical tie, and announced a growing EU tier requiring EU-registered companies with EU-located GPUs.

    The emissions fix and reversed ruling directly affect miner payouts and standing, and the new EU tier changes eligibility requirements going forward.

  4. 04

    SN6 (Numinous) launched an open-ended forecasting competition where miners predict the future state of a dynamic knowledge graph, replacing the subnet's prior Polymarket-style prediction mechanism, which the team paused rather than removed.

    It is a full mechanism change on an active subnet, with the team confirming existing hotkeys carry over and that it cannot maintain both competitions at once.

  5. 05

    A community member reported a regression test tied to the Yuma correction that funds 133.2 alpha for miners but allocates only 120.6, now included in runtime 482 on Finney.

    A discrepancy between funded and allocated alpha in a consensus-related correction could affect miner payouts if not addressed.

Subnet Community Pulse

5

SN23 24h +28.02%

Trishool

What happened
SN23 (Trishool) launched Astroware, a separate commercial entity under the same founding team that buys continuous red-teaming work from the subnet under a commercial agreement. Astroware introduced HaloGuard EU, a product that uses Trishool's work to help align model outputs with EU compliance requirements. The subnet team also described several banned miner submission techniques it has identified, including safety-prefix priming and quoted-payload isolation.
Community
Community members shared the Astroware and HaloGuard EU announcements and reacted positively, with one member calling it an 'awesome team.' A miner asked for clarification on whether special scripts are needed to avoid having submissions flagged under the new detection rules, and another asked when the burning rate would decrease.
Official
The team posted an 8-part thread explaining that Trishool will receive a share of Astroware's product revenue, with the exact share still being set, and that revenue will initially fund alpha buybacks intended mainly for burning. The team said buybacks have not started yet since the product just launched, and that an on-chain dashboard will be published once they begin.

SN15 24h −1.10%

ORO

What happened
SN15 (ORO) switched its ORO Bench to daily dynamic "worlds" for each race, pausing submissions and resetting all agent scores to prevent hardcoding and overfitting. Submissions were reopened shortly after the switch. Separately, several miners reported that newly submitted agents were incorrectly flagged by the anti-cheat system, and the team pushed updated static-analysis checks and reset cooldowns for the affected hotkeys.
Community
Many miners said the new qualifying score threshold is too high and the qualifying time window too short, repeatedly asking the team to lower the threshold or extend the window. A few participants instead described the reset as a fresh opportunity for newer miners to compete against previously dominant agents.
Official
The team said flagged agents found to be false positives would be pushed through for qualifying, updated static-analysis checks, and stated that qualifying thresholds would not change because all miners had the same amount of time.

In debate · SN15 ORO Bench daily-worlds reset backlash · 55 participants

  • Miners argue the post-reset qualifying threshold is unfairly high and should be lowered.

    “The threshold for qualifying in the new benchmark is too high.”

  • Some community members see the reset as a fair new chance for less established miners.

    “it's a new chance for newbies, like me”

  • The team maintains the rules are equal for everyone and declines to adjust the threshold.

    “Everyone had the same amount of time to work on their agents for todays race”

SN92 24h −9.99%

REAX

What happened
SN92 (REAX) fixed a validator bug that had zeroed emissions for several engine-track UIDs after an outdated engine state file was read. A statistical tie among uid 250, uid 116 and uid 199 in the generalist opening playoff led the team to reverse its initial single-champion decision and name all three co-champions. On the engine track, uid 56's champion award was held because its kernel binaries lacked published source, then approved after the code was released under an MIT license; a new engine champion, uid 176, was crowned afterward. The team also announced a growing EU tier requiring miners to be EU-registered companies with EU-located GPUs, published a public miner FAQ, and set fixed biweekly crowning slots for the generalist track.
Community
Several miners and a validator pressed the team for the underlying numbers, arguing that uid 116's higher playoff score should have decided the generalist opening rather than commit order, while others questioned why only three entries received an extension draw when additional entries were within the same noise level. Separately, some miners reported that mining has become unprofitable across various GPU types.
Official
The team published the full playoff scores and standard errors, reversed the initial sole-champion call in favor of three co-champions, and later detailed the exact score margin used to define the statistical tie group.

In debate · SN92 generalist playoff tiebreak fairness · 18 participants

  • The team initially applied its published rule that the earliest commitment among statistically tied entries wins the opening.

    “the published tie rule applies: the earliest commitment among the tied entries wins”

  • Community members argued the champion should be decided purely by score since all entries competed as one cohort.

    “the champion should simply be the model with the highest score”

  • Other miners questioned why only three entries were treated as tied when several more scored within the same noise level.

    “all are within noise level. what's special in those 3 co-champions and what's wrong with the other 3 losers?”

SN6 24h −1.17%

Numinous

What happened
SN6 (Numinous) launched an open-ended forecasting competition in which miners predict the future state of a dynamic knowledge graph, starting with top US equities and their supply chains and adding a geopolitics-focused graph the following day. The new mechanism replaced the subnet's prior Polymarket-style prediction competition, which the team paused. Validator and miner setup code and documentation were released alongside the launch, and public API endpoints for leaderboard and board data were published ahead of a dedicated graph-competition leaderboard.
Community
Miners asked whether existing hotkeys could be reused for the new competition and whether the old Polymarket-based agents needed to be fully replaced or could continue running. One miner said the switch was disappointing after spending days developing a forecasting agent for the paused competition just before receiving its first scores. Other questions covered round timing, immunity periods, and whether multiple UIDs could be used to forecast.
Official
The team confirmed existing hotkeys can be reused for the graph competition, that the Polymarket-based mechanism is paused rather than removed, and that the move was necessary because the team cannot maintain both mechanisms at once. It also noted the graph competition's move outside the sandbox had been announced in early October.

SN105 24h −2.45%

Beam

What happened
On SN105 (Beam), a miner raised concerns about the fairness and transparency of room-transfer eligibility, questioning why some requests were approved while others were not. Other miners argued that the paid room-transfer economics could let capital outweigh engineering skill in determining emissions. Separately, a miner built and demonstrated a live video and audio streaming proof-of-concept over Beam's messaging protocol, running it through a Discord-hosted room-viewer tool.
Community
Miners debated the allocation dispute at length, with one pointing to the publicly available assignment-plan data showing allocation tied to PRISM scores, while others continued to question the broader incentive design. Separately, miners discussed creating a dedicated community group to help increase Beam's visibility.
Official
The team stated that streaming and message protocols are currently excluded from PRISM scoring, so miners can test without penalty, and offered an unlimited API key for streaming cost testing. The team also praised the miner's streaming demo.

More subnet community coverage every day in the Telegram channel →

Voices on X

  • @cascade_sn91

    SN91 (Cascade) said it is preparing Cascade-MoE, a mixture-of-experts time-series model targeting over 1B+ parameters, up from its earlier 4M-parameter model.

    View the original post on X ↗

  • @TaoOutsider

    TaoOutsider reported a regression test tied to the Yuma correction that funds 133.2 alpha for miners but allocates only 120.6, now included in runtime 482 on Finney.

    View the original post on X ↗

  • @SubnetSummerTAO

    SubnetSummerTAO said SN68 (NOVA) compounds showed experimental ADME results exceeding reference compounds, including approved drugs, as discussed at the Exploit Summit.

    View the original post on X ↗

  • @trishoolai

    Trishool said it launched a new product this week connecting Astroware and Trishool within SN23, describing how value flows back to the subnet.

    View the original post on X ↗

  • @TaoOutsider

    TaoOutsider reported that under eTAO, SN41 (Almanac) cut its owner allocation from 18% to 10% while SN8 (Vanta) set it to 0%, with SN36 and SN70 at 33%.

    View the original post on X ↗

  • @forevermoney_ai

    forevermoney_ai said SN80 will launch on Solana via Chainlink CCIP, becoming the first Bittensor subnet live on three chains: Base, Robinhood Chain and Solana.

    View the original post on X ↗

  • @mvtrx_79

    mvtrx_79 said its September 28 scoring overhaul more than doubled liquidity, left aggressive order volume down three quarters, and cut emissions to gaming registrations to a few percent.

    View the original post on X ↗

  • @Bitcast_network

    Bitcast_network said $93k was bought back over the last 30 days, funded by companies paying to reach audiences across the Bitcast network.

    View the original post on X ↗

◆2026-10-23 · End of this edition

Protocol

6 PRs

#3276 Breaking change

Fix: allow root claims for baskets holding all 128 subnets plus TAO

What
The pull request raises the admission limit for root claims per hotkey from 129 to 130 so that a basket covering all 128 non-root subnets plus root (TAO) can be claimed.
Why
The previous limit excluded the one basket configuration that includes every subnet plus TAO from being eligible for a root claim.
Who it affects
This affects TAO holders who hold a basket spanning all subnets plus TAO and want to claim root; it is a breaking change tied to a runtime spec version update, so node operators and validators need to run the updated version.

#3101 Breaking change

Add continuous stake reporting throughout subnet dissolution lifecycle

What
The pull request adds tracking of a subnet's lifecycle state so that subnets going through dissolution, and their stake, remain visible in reporting data, while stake that has already been settled is hidden from the report to avoid counting it twice.
Why
Reporting previously could lose visibility into stake during subnet dissolution or double-count stake that had already been settled.
Who it affects
This affects validators and TAO holders who rely on stake reporting data during a subnet dissolution; it is a breaking change for anyone consuming that reporting data.

#3102

Add liquid alpha consensus mode controls to precompiles

What
The pull request adds functions to the subnet precompile that let EVM-based tools read and change the Liquid Alpha Consensus Mode setting.
Why
This gives smart-contract and EVM tooling direct, typed access to a setting that previously required other interfaces.
Who it affects
This mainly affects subnet owners and developers who manage subnet settings through EVM tools; ordinary TAO holders, miners, and most validators are not affected.

#3225

test(runtime): add ±20% fee regression snapshots and EVM transaction coverage

What
The pull request replaces the fee test data set with a larger, more detailed set of test cases and adds tests for signed transactions covering transfers, staking, and proxy actions.
Why
The change aims to catch unintended fee changes before they reach production by checking that fees stay within a defined range of previous values.
Who it affects
This is a testing and development change only; it does not alter runtime behavior, fees, or interfaces, so no one outside the development team is affected.

#3253

Fix stale token symbols in metagraph displays

What
The pull request fixes the metagraph display in the CLI so it shows the correct, current token symbol instead of a stale cached one for alpha balances and emissions.
Why
A caching issue could cause the wrong currency symbol to be shown, particularly when viewing data from around the time a subnet's symbol changed.
Who it affects
This affects TAO holders and others viewing metagraph data through the CLI; it is a display-only fix with no change to underlying amounts or runtime behavior.

#3277

Fix btcli claim-root crash when --subnets is omitted

What
The pull request fixes a crash in the btcli claim-root command that occurred when the --subnets option was left out.
Why
A coding error in how the CLI generated default option values caused the command to fail instead of applying its intended default.
Who it affects
This affects users of the btcli claim-root command who omit the --subnets option; it is a CLI-only fix with no runtime or protocol-level effect.

Recent upgrades

Recent editions

Past editions +
2026-10-232026-10-222026-10-212026-10-202026-10-192026-10-182026-10-172026-10-162026-10-152026-10-142026-10-132026-10-122026-10-112026-10-102026-10-092026-10-082026-10-072026-10-062026-10-052026-10-042026-10-032026-10-022026-10-012026-09-302026-09-292026-09-282026-09-272026-09-262026-09-252026-09-242026-09-232026-09-222026-09-212026-09-202026-09-192026-09-182026-09-172026-09-162026-09-152026-09-142026-09-132026-09-122026-09-112026-09-102026-09-092026-09-082026-09-072026-09-06