THE DAILY BRIEFING

Bittensor Daily — October 28, 2026

Protocol work dominated the day's substance: a new runtime release brings quantum-resistant wallets and tighter proxy controls, while #Rao kept chewing on two open tokenomics questions -- a new 'emission decay' proposal and the lingering debate over near-zero subnet registration fees -- without any official reply. On the subnet side, SN78 (Umi) closed out its Cohort 6 scoring dispute and SN60 (Bitsec.ai) confirmed a benchmark oversight. #General stayed mostly procedural, repeating its standing scam warning and walking a member through a staking mechanic.

A member proposes 'emission decay' in #Rao, tying a mature subnet's continued emissions to value it actually returns rather than to demand for its alpha
How will Bittensor incentivise subnets to return more value than they receive in emissions? Emission decay would gradually reduce a subnet's emissions over time, but subnets could slow or reverse that decay by returning value to alpha holders (1) The proposal uses Lium's flywheel purely as an illustration, without alleging wrongdoing, to flag a structural incentive gap: rising TAO price could scale subsidies to mature subnets faster than their revenue justifies. (2) It is a distinct angle from the week's earlier tokenomics threads -- Oct 20's bonding-curve proposal and Oct 26's falling-registration-price debate -- since it targets ongoing emissions rather than entry or exit costs. (3) As with those earlier threads, no official or core-dev voice responded, leaving the mechanism's feasibility unaddressed in the material.
The Oct 26 debate over Bittensor's falling subnet registration fee gains a counterargument: cheap entry will not keep small teams alive after launch
low SN reg fee is the most obvious purely positive change ever Lowering the subnet registration fee is the best way to attract real, new innovation. yea sure will help but still no chance to survive/shine after launch without whales behind you (1) The thread directly continues Oct 26's #Rao item on registration costs falling toward near-zero, which that day was treated mostly as a mechanical curiosity. (2) Today's exchange adds a distinction not present earlier -- that a low entry fee helps small teams register but says nothing about their odds of surviving against better-funded competitors once live. (3) No official or core-dev voice engaged with either side, consistent with the pattern across the week's #Rao tokenomics threads.
Subtensor ships Release 487, adding quantum-resistant wallet key types and finer-grained proxy permissions
(1) Release 487 raises the runtime's spec_version from 486 to 487 to accompany support for quantum-resistant wallet types -- ML-DSA-65, Falcon-512 and Falcon-1024 -- plus an optional key rotation feature, while existing SR/ED wallets continue to work unchanged. (2) Separate changes landed alongside it: a new lock proxy type that can manage stake lock conviction periods without being able to move, add or remove stake, and a rule letting staking-proxy holders claim root dividends directly with one proxy permission instead of two. (3) The wallet and versioning changes are breaking for node operators, who must update, but optional for TAO holders, validators and miners who do not use proxies or choose not to adopt the new key types.
SN78 (Umi) resolves its Cohort 6 scoring dispute, deciding to score only the eight new evaluation clips rather than the original six or a blended set
(1) The decision follows the memorization problem disclosed earlier this week, in which several top-scoring Cohort 6 models had gamed evaluation by memorizing public test data rather than generalizing. (2) The owner set UID 172's model as the Cohort 7 baseline and shifted the subnet's ETAO emission split to move incentive from validators toward miners and the owner, while also adjusting the per-bucket emission decay used from Cohort 7 onward. (3) C6 weights went live on-chain and Cohort 7 submission intake opened with a planned 12-hour closing warning, even as reward release to validators remained pending; the subnet's token rose 8.0% over 24 hours.
On SN60 (Bitsec.ai), the team confirms a benchmark oversight: a vulnerability the subnet is supposed to test for had already been patched in the evaluation image
(1) A community member reported that Initia Move's H-02 vulnerability, which the SN60 benchmark tests for, had already been patched, meaning agents could not legitimately detect it in that round. (2) The team confirmed the patch had gone unnoticed, calling it an oversight, while the round continued to show four errors; it also clarified that evaluation project keys are never disclosed in advance and can change between rounds, visible to miners only during evaluation. (3) The reporting member asked for a re-screening of agents evaluated after the report was made; no re-screening outcome appears in the material.
Other: #General answers a root-stake question and reruns its scam warning, while #Rao leaves a subnet-ownership query unresolved
you do need to have some free TAO on the coldkey in order to unstake. (1) A moderator resolved a member's confusion over an unclaimed root balance by pointing to the free-TAO requirement, while the channel's standing imposter-scam warning ran again unchanged. (2) In #Rao, a moderator redirected a question about a deregistered owner's subnet status to the help forum rather than answering it directly, leaving the ownership question open.

#General

2 REPORTS

Notable

A member's root-stake claim fails until a moderator explains that unstaking requires free TAO on the coldkey

Opening · 09:23 UTC

A member asked I have 0.4 Tao to claim In root but it doesn’t work, do I need to have x amount of free Tao or what’s the matter (all is staked via Tao bot and on chutes 50/50).

Reply · 14:00 UTC

A moderator pointed to the FAQ and confirmed the requirement directly: you do need to have some free TAO on the coldkey in order to unstake.

2 things to read from it

The friction is procedural rather than technical: a balance that looks claimable on-chain is blocked by a free-TAO requirement the asker did not anticipate, a gap the FAQ apparently exists to cover but that still reaches the channel.
This sits alongside Oct 26's detailed #General thread on liquidation-price mechanics as another case where ordinary staking mechanics prove non-obvious enough to require a moderator's manual explanation rather than being self-evident from the interface.

Where it stands

Answered directly by a moderator with a link to the existing FAQ entry; no further follow-up from the asker in the material.

Other

Other: a moderator redirects subnet-specific complaints out of #General, a newbie dispute flares briefly, a findings report is routed to a thread, and the channel's standing scam warning runs again

Moderation · 00:44–00:45 UTC

A moderator told the channel Hey don't clog general chat with internal subnet issues and added that members should report any issue you might be having in their subnet channel and handle it there.

Dispute · 00:00–00:01 UTC

Earlier, one member told another hey, you're a newbie. you should show some respect to your seniors. while a second member asked what happened if your warning is ignored?.

Routing · 04:31–05:50 UTC

A member asked to share findings privately with Hey [mention] , may i send you a dm regarding my findings?, and the moderator instead pointed to a dedicated thread, saying You are free to send your findings in. A <#1107738550373454028> thread and later I have responded to your thread, while separately reminding another member Please make your profile public and keep it that way. Private profiles are not allowed in this server.

Scam warning · 13:59 UTC

A moderator reposted the channel's standing imposter warning in full, stressing Anyone DMing or calling you right now is a scammer, even if they look like me, another mod, admin, Opentensor Foundation team member, etc (they are imposters). We will not send you a DM or FR.

2 things to read from it

Volume held roughly steady at 14 messages, continuing the drop from Oct 26's spike of 27, with most of today's traffic again being moderation and routine redirection rather than ecosystem substance.
The repeated scam warning and the private-profile reminder echo the standing moderation pattern already noted on Oct 26, reinforcing that these are recurring, not new, friction points.

In the week’s context

On Oct 26, #General logged a money-mule scam caught and banned alongside the channel's standing imposter-scam reminder, the same template repeated again today.

Where it stands

All items closed within the day through moderator action or routing; no open ecosystem question remains unanswered.

#Rao

3 REPORTS

Key

A member proposes 'emission decay' as a mechanism to tie subnet funding to value actually returned to the network

Proposal · 22:52 UTC

A member opened with a direct question -- How will Bittensor incentivise subnets to return more value than they receive in emissions? -- and used Lium's flywheel to illustrate the risk: Take roughly $60m in annualised emissions, $15m ARR and 100% of revenue going towards buybacks Those buybacks support alpha's price and can help attract more emissions.

Mechanism · 22:52 UTC

The same member sketched a scaling concern if TAO's price rises -- Now imagine TAO rises 4x to $1,000 With the same TAO allocation, that $60m annualised subsidy becomes $240m -- and proposed a fix: Emission decay would gradually reduce a subnet's emissions over time, but subnets could slow or reverse that decay by returning value to alpha holders.

Rationale · 22:52 UTC

They framed the goal as a maturity test for subnets: As subnets mature, they should justify continued support from the network by giving value back Those that don't would see their emissions decline, adding that the approach would also let the network dynamically adjust emissions as TAO's price f changes.

3 things to read from it

This is a new tokenomics angle distinct from the LP-curve and bonding-curve debates seen earlier in the week -- rather than addressing registration or deregistration costs, it targets the ongoing relationship between a mature subnet's revenue and its continuing emission allocation.
The proposal uses a real subnet (Lium) as its working example without alleging wrongdoing, framing the concern as a structural incentive question rather than a criticism of that subnet specifically.
As with the LP-curve, shorting and OHM-comparison threads logged on Oct 20 and Oct 26, the post arrived with no reply from any official or core-dev voice in the material, leaving the idea to stand alone for now.

In the week’s context

The proposal extends a week of unanswered tokenomics proposals in #Rao, including Oct 20's bonding-curve discussion (PR3211) and Oct 26's falling-registration-price debate, none of which have drawn an official response.

Where it stands

Posted without reply in the material; whether emission decay is technically feasible or under any consideration remains unaddressed.

Notable

The low subnet-registration-fee debate from Oct 26 continues, with a new pushback that cheap registration alone will not help small teams survive

Continuation · 09:05–09:07 UTC

A member restated the earlier position bluntly: low SN reg fee is the most obvious purely positive change ever, after an exchange that opened with can you make a respect from degenerates? and why you guys make jokes.

Support · 16:22 UTC

Another member agreed and extended the argument toward who would benefit: we should be attracting small teams with limited funding, like grad students and academic researchers. Bitstarter isn't the answer either, it helps, but introverted tech nerds don't want to present a CV to a board. Lowering the subnet registration fee is the best way to attract real, new innovation.

Pushback · 19:03 UTC

The original poster of the agree-chain then qualified their own earlier enthusiasm, replying to that support with yea sure will help but still no chance to survive/shine after launch without whales behind you.

3 things to read from it

This is the same registration-price debate flagged in Oct 26's edition, but today it gains a substantive counterpoint: a lower entry fee may help small teams register, but does not address what happens to them after launch without large backers.
The pushback introduces a distinction not present in Oct 26's framing, which treated the falling price mostly as a mechanical curiosity -- today a participant ties it explicitly to post-launch survivability.
No official or core-dev voice engaged with either side of the exchange, consistent with the pattern across the week's #Rao tokenomics threads.

In the week’s context

Directly continues Oct 26's #Rao item on the subnet registration price falling toward near-zero, which was then treated half-jokingly by members with no official reply.

Where it stands

Open; the tension between cheap registration and post-launch viability was raised but not resolved or addressed by any team or protocol voice.

Other

Other: a question about a deregistered owner's new subnet is redirected to the help forum

Exchange · 16:37–19:02 UTC

A member asked a moderator directly, I want to know which subnet [mention] owns. I think his all subnets were deregged. Did he register new subnet? Or did you forget to remove his owner role?, and was told Please do not tag me again for this. Just post in the help forum and let any moderator help you with the problem., drawing a reaction of XDDD insane from another member.

What to read from it

A routine moderation redirect rather than a substantive answer; the underlying ownership question was not resolved in the channel.

Where it stands

Closed by moderator redirect; the ownership question itself remains unanswered in the material.

Subnets

5 REPORTS

SN112 24h −1.41%

THOR

What happened: SN112 (THOR) launched as a new robotics-data subnet on Bittensor, built around turning robotics footage into training data for robotics labs and researchers. The initial task is action captioning on egocentric (first-person) video, with the subnet currently in a private beta using whitelisted miners. The team said GitHub access and a public application process would follow as the beta progresses.
Community: Community members asked about the purpose of the revenue dashboard, the high registration fee, and how to apply as annotators, including a team describing itself as a large annotation group based in Brazil. One community member suggested that if the model works as intended, the subnet could become the first Bittensor subnet in robotics to generate real revenue, and another noted that high deposits are common when robotics labs commission this type of work.
Official: The SN112 (THOR) team said the revenue dashboard tracks signed-contract deposits and buybacks, and that the current registration fee is set high deliberately during the private pilot phase, with a likely reduction after the beta. The team also said annotation work may be AI-assisted, human-annotated, or a mix of both depending on the task, and that wider participation will open once beta quality is validated with clients.

SN78 24h +8.00%

Umi

What happened: SN78 (Umi)'s owner resolved an extended debate over C6 scoring by deciding to use only the eight new evaluation clips rather than the original six, the combined 14, or a 50:50 split, with UID 172's model set as the baseline for C7. C6 scoring and certification were completed and corrected results published, though reward release to validators was still pending. C6 weights later went live on-chain, C7 submission intake opened with a planned 12-hour closing warning, and the subnet's ETAO-based emission split was changed to shift incentive from validators toward miners and the owner (owner 18→30.8%, validators 41→15.5%, miners 41→53.8%). The owner also said the per-bucket emission decay for C7 and later cohorts would move from 0.75 to 0.85, with a 1% grace margin added around the baseline.
Community: Miners debated whether the scoring change was fair to overfitted versus non-overfit models, with some calling the shifting rules confusing and others supporting the move to the new clips. A miner also proposed giving the top 3-5 models fixed, score-ordered reward shares to reduce large payout swings between closely scored models.
Official: The owner confirmed the final C6 scoring table would use the new 8 clips, announced the ETAO split and C6 weights going live, and outlined the C7 emission decay and grace-margin adjustments.

In debate · SN78 C6 scoring methodology dispute · 5 participants

Proponents argued the new 8 cross-dataset clips better measure non-overfit performance, since models trained on the original six scored poorly on them.

“We’d support using only the eight new clips for C6, since the 14 models that scored ≥80% on the original six average just 32.5%”
A miner proposed a 50:50 split between the combined 14-clip table and the new 8 to balance the certified C6 suite with cross-dataset results.

“I suggest a compromise idea for C6: split the model pool 50:50, half allocated by the combined 14-clip table and half by the new 8”
Another miner warned that repeated changes to the scoring rules created uncertainty for miners who plan around announced decisions.

“the C6 rules changed several times (bands, formula, then 14 → 50:50 → new 8), and during the endpoint cohorts too”

SN103 24h +30.70%

Terza Inference

What happened: SN103 (Terza Inference) launched on mainnet, providing access to miner-provided inference models through a single API key. Separately, the subnet's Discord channel, which had been unclaimed, was claimed by a new owner; moderators had previously warned against posting unverified ownership claims sourced from other platforms.
Community: Community members discussed whether mining had started while the channel remained unclaimed, with one noting no owner had yet come forward. After the claim was announced, reactions included surprise at the speed of the process and a question about who was currently receiving emissions.
Official: A moderator confirmed the channel had been successfully claimed and asked members to wait for informational posts to be pinned before making further requests.

SN58 24h +18.60%

Attune

What happened: SN58 (Attune) miners reported that spam model submissions were gaming the subnet's low scoring threshold, with several spam models reaching top ranking positions above legitimately trained models. The team acknowledged the issue in the subnet channel and said it is working on improvements tied to an upcoming "level 3" update. Separately, a community member proposed an escalating per-submission fee mechanism, modeled on SN49 (Tensorplex?)'s approach, to deter further spam.
Community: Miners expressed frustration that spam models were outranking trained submissions, calling the situation unfair to legitimate miners, while some reacted with dark humor about the 2.9 scoring threshold. A proposal to add a registration fee alongside an escalating submission fee drew pushback from miners who said the current registration fee is already high relative to low emissions.
Official: The team acknowledged the spam reports, asked miners for patience, and said it is working on scarce resources ahead of a planned "level 3" update.

In debate · How to stop spam model submissions from topping the SN58 (Attune) leaderboard · 9 participants

Miners said spam submissions unfairly overtook legitimately trained models under the current low threshold.

“Take a look at how spam models become the best 🤣. 2.9-score miner will be so sad.”
The team said it is constrained by scarce resources and asked for patience ahead of planned improvements.

“We’re working on scarce resources for now. Our team currently working very hard for the next level, improvement and everything.”
A community member proposed an escalating per-submission fee, similar to SN49's mechanism, to discourage spam.

“I think SN49's mechanism could be an effective way to prevent spam submissions.”

SN60 24h −5.81%

Bitsec.ai

What happened: A community member reported that the Initia Move project's H-02 vulnerability, which the SN60 (Bitsec.ai) benchmark is supposed to test for, has already been patched in the current benchmark image, meaning agents cannot legitimately detect it; the member requested a re-screening of agents evaluated after the report. Separately, miners asked whether the evaluation project list for a round is disclosed at submission time, and the team clarified that project keys are not known in advance and can change between rounds. A further discussion traced how one participant obtained a project list from hard-steer check logs, revealing that the project set had changed at the last moment before a round started.
Community: Several miners supported the request for re-screening, with one independently verifying the H-02 patch claim against the benchmark source code. Miners also pressed the team with detailed questions about when and how round-specific project lists become visible to agents, and disputed the source and timing of a leaked project list.
Official: A team member confirmed the H-02 vulnerability in Initia Move had been patched while the round still showed four errors, calling it an oversight. The team also confirmed that project keys are not disclosed in advance and can change between rounds, visible to miners only during evaluation.

More subnet community coverage every day in the Telegram channel →

@TargonCompute

Targon (SN4) said it repurchased and burned 1,772 TAO (~$445K) and a further 6,678 TAO (~$1.7M) of SN4 alpha, funded by organic protocol revenue.

View the original post on X ↗
@chutes_ai

SN64 (Chutes) expanded its catalog with new non-TEE models, including DeepSeek V4 Pro, Qwen3.8 Max, GLM-5.2-Fast, Qwen3.8-2.4T-A95B, and GLM-5.3, alongside its existing TEE models.

View the original post on X ↗
@SynthdataCo

SN50 (Synth) said its open-sourced AutoQuant tool, giving AI agents $100 and 18 hours to build a forecasting model, has not yet beaten its own live quant miners.

View the original post on X ↗
@yanez__ai

SN54 updated its eTAO emission split, raising validators' share to 49% from 41%, cutting miners' share to 33% from 41%, and ending the 30% miner burn.

View the original post on X ↗
@TaoOutsider

SN116 (Carbon), built by BitStarter AI and Carbon Physics AI, is approaching mainnet launch, according to a community account.

View the original post on X ↗
@taodaily_io

A post said IOTA SDK produced nearly identical AI training behavior across seven runs of about 140 GPUs each, despite 585 GPU crashes, preemptions, and rejoins.

View the original post on X ↗
@TaoOutsider

A community member flagged open PR #3285, which would let Bittensor skip the roughly day-long wait to reassign stake weight when an unavailable child validator is unregistered, lacks a permit, or is inactive.

View the original post on X ↗
@TerzaInference

SN103 (Terza) launched on mainnet, offering access to multiple miner-provided inference models through a single API key.

View the original post on X ↗

◆2026-10-28 · End of this edition

#3293 Breaking change

Add lock proxy type

What: Subtensor adds a new proxy type that can manage stake lock conviction periods but cannot move, add or remove stake.
Why: The existing proxy types did not offer a way to delegate only the management of stake locks without also granting broader staking permissions.
Who it affects: Validators and miners who use proxy accounts can now assign a narrower permission set for lock management; it has no direct effect on ordinary TAO holders who do not use proxies.

#3295

Allow root claims through staking proxies

What: Accounts holding a staking proxy can now claim root dividends directly, without needing a separate root-claim proxy permission.
Why: The change removes the need to hold two separate proxy types to perform a single related action.
Who it affects: Validators and TAO holders using staking proxies can now claim root dividends with one proxy permission instead of two; miners are not affected.

#3220

Add quantum-resistant wallets and key rotation

What: Subtensor adds support for quantum-resistant wallet key types (ML-DSA-65, Falcon-512, Falcon-1024) and an optional key rotation feature, alongside the existing SR/ED wallets.
Why: The addition gives users an alternative key format with stronger resistance to quantum attacks, without changing current wallet behavior.
Who it affects: TAO holders, validators and miners may optionally adopt the new wallet types or key rotation; existing wallets continue to work unchanged.

#3291 Breaking change

Release 487 - Quantum Resistant Wallets

What: This release raises the runtime's spec_version from 486 to 487 to accompany the quantum-resistant wallet changes.
Why: A spec_version bump is required whenever runtime logic changes so that nodes recognize the new version.
Who it affects: This is a technical versioning step relevant mainly to node operators and developers; it has no direct effect on ordinary TAO holders, validators or miners beyond requiring updated nodes.

Past editions +

24H Market

Gainers

Losers

Bittensor Daily — October 28, 2026

#General

A member's root-stake claim fails until a moderator explains that unstaking requires free TAO on the coldkey

Other: a moderator redirects subnet-specific complaints out of #General, a newbie dispute flares briefly, a findings report is routed to a thread, and the channel's standing scam warning runs again

#Rao

A member proposes 'emission decay' as a mechanism to tie subnet funding to value actually returned to the network

The low subnet-registration-fee debate from Oct 26 continues, with a new pushback that cheap registration alone will not help small teams survive

Other: a question about a deregistered owner's new subnet is redirected to the help forum

Subnets

THOR

Umi

Terza Inference

Attune

Bitsec.ai

Add lock proxy type

Allow root claims through staking proxies

Add quantum-resistant wallets and key rotation

Release 487 - Quantum Resistant Wallets

24H Market

↗Gainers

↘Losers

Bittensor Daily — October 28, 2026

Today's highlights

Discord conversations worth following

A member's root-stake claim fails until a moderator explains that unstaking requires free TAO on the coldkey

Other: a moderator redirects subnet-specific complaints out of #General, a newbie dispute flares briefly, a findings report is routed to a thread, and the channel's standing scam warning runs again

A member proposes 'emission decay' as a mechanism to tie subnet funding to value actually returned to the network

The low subnet-registration-fee debate from Oct 26 continues, with a new pushback that cheap registration alone will not help small teams survive

Other: a question about a deregistered owner's new subnet is redirected to the help forum

THOR

Umi

Terza Inference

Attune

Bitsec.ai

Voices on X

Protocol

Add lock proxy type

Allow root claims through staking proxies

Add quantum-resistant wallets and key rotation

Release 487 - Quantum Resistant Wallets

Recent editions

Gainers

Losers