Daily digest of the Bittensor ecosystem

中文/EN

Daily edition
Nov 11, 2026

Bittensor Daily A 1T1B product
1T1B ↗
BITTENSOR DAILY

24H Market

01/10

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

THE DAILY BRIEFING

Bittensor Daily — November 11, 2026

The day's substance ran mostly through the protocol and the subnet cards rather than through sustained Discord debate: Opentensor's Gamma rollout dominated the X feed, the multi-day root-basket deposit freeze finally produced a concrete fix pointer, and SN36 (Weir) and SN124 (Swarm) each drew unresolved questions about who controls emissions and seats. Discord itself stayed thin -- 24 messages in #General and 3 in #Rao -- with routine moderation, a resolved collateral-mechanics question, and a scam-offer pattern spreading across multiple subnet channels filling the rest of the day.

Today's highlights

  1. Opentensor formally launches Gamma Tokens, a native payment system letting Bittensor wallets pay directly for subnet services without separate accounts or API keys.

    Introducing Gamma Tokens A native payment system for Bittensor’s AI economy. Users and AI agents will be able to access and pay for subnet services with a Bittensor wallet, without separate accounts or API keys for each subnet. Gamma is live on Bittensor. It was enabled last night. How do you find out which subnets have enabled gamma? How much has been minted/spent? (1) The mechanism burns a subnet's alpha to mint dollar-valued credit, with Creating Gamma always burns the subnet’s alpha, permanently removing those tokens from circulation tying service payments directly to supply reduction. (2) This follows the Gamma credit mechanism merging in Release 488 on Nov 5 and sitting live but switched off on Nov 6; it is now formally announced even as taostats itself asks which subnets have actually opted in. (3) Subtensor separately shipped a related gamma-emission PR today letting subnet owners direct emission shares to buy and burn another subnet's alpha, extending the same mechanism at the protocol level.

  2. The root-basket deposit freeze running since Nov 7 gets its first concrete fix pointer, offered by a community member rather than the team.

    A member still waiting on the fix asked When will this be fixed? The balance in my basket still hasn't changed., and another replied its in the release branch. My guess is it will go out tomorrow. Separately in #General, a member wrote I think rewards are back, prompting the reply Hopefully we'll get an upgrade announcement in the <#830075335084474390> channel. (1) This is the first time in four days that either channel has pointed to an actual artifact rather than repeating that a fix is being worked on, following Nov 7's confirmed deposit block and Nov 8's design critique of the blocking gate. (2) The two channels again describe the situation without cross-referencing each other, with #General suggesting rewards may be returning while #Rao still waits. (3) No moderator or team voice confirmed the pull request or gave a firm release date.

  3. SN36 (Weir) shows active mining and roughly 20 TAO/day in emissions despite having no confirmed owner in its official channel, even as its token rose 52.91%.

    (1) Moderators said on-chain identity fields can be set arbitrarily by whoever holds the key and called such sources low-confidence, declining to confirm who controls the subnet. (2) No individual or team has stepped forward to claim ownership or explain the emissions, leaving the question of who is extracting roughly 20 TAO a day unresolved. (3) Moderators directed emissions-related questions to the foundation, noting they are not OTF or Rao Foundation employees, leaving the subnet's governance status open even as its price moved sharply.

  4. SN124 (Swarm) faces allegations of seed leakage behind its top model and coldkey concentration across Sentinel seats, denied by the team the same day it announced a new buyback program.

    20 TAO buyback & burn was one of four changes the subnet announced, alongside restricting output to paying clients and increasing staker allocation. (1) A miner separately alleged the leading model may have had evaluation seeds in advance, pointing to large score swings for the top-ranked UID, while a community report alleged one coldkey controls roughly 60% of Sentinel seats through multiple hotkeys. (2) The team said it reviewed for seed leakage and found none, attributing the swings to overfitting, and declined to make Sentinel scoring fully public, citing a client confidentiality agreement. (3) The subnet's token rose 4.14% over 24 hours despite the unresolved allegations.

  5. Other: a single account posts identical subnet-slot purchase offers across at least ten subnet channels, while #General resolves a collateral-mechanics question and logs routine scam warnings.

    (1) The templated offers appeared in at least ten separate subnet channels, drawing attention from community members and moderators without a confirmed resolution in the material. (2) In #General, a member's question about where locked registration collateral goes when a subnet is pruned was resolved with a documented answer: If the subnet dissolves, the lock metadata is cleared after alpha is settled to coldkey free TAO with everyone else's stake — there is no separate collateral refund step. (3) The channel otherwise ran its standing scam-imposter warning and routine moderation, continuing the week's pattern of low ecosystem substance in #General and #Rao.

Discord conversations worth following

#General

4 REPORTS

Notable

The root-basket deposit freeze, running since Nov 7, shows its first signs of a fix landing as a member reports rewards returning and a #Rao contributor points to a pull request already in the release branch.

Opening ·

A member in #General wrote I think rewards are back, prompting a moderator to say Hopefully we'll get an upgrade announcement in the <#830075335084474390> channel.

Parallel ·

In #Rao, a member still waiting on the fix asked When will this be fixed? The balance in my basket still hasn't changed., to which another member replied its in the release branch. My guess is it will go out tomorrow.

3 things to read from it

  1. This is the fourth straight day the deposit freeze has surfaced in these channels, but the first day either channel points to a concrete fix location rather than only the generic known-issue line used since Nov 7.
  2. The two channels again describe the same situation without cross-referencing each other: #General has a member saying rewards appear to be returning while #Rao still has one waiting on a fix, with no official announcement in either channel confirming either account.
  3. No moderator or team voice confirmed the pull request or gave a firm timeline; the 'tomorrow' estimate came from a community member, consistent with the pattern across this week's freeze reports.

In the week’s context

  • Nov 7: the team confirmed Root dividends were accruing but blocked from entering stakers' baskets, with no fix timeline given.
  • Nov 8: a #Rao member argued the blocking gate was doing more harm than the bug it guards against, a critique that went unanswered.
  • Nov 9: a member reported a validator's entire basket appearing to vanish after an unstake, the third consecutive day the issue surfaced in #General with no fix timeline.

Where it stands

Still open: a fix is apparently queued in the release branch per a community estimate, but no official upgrade announcement had appeared in either channel by day's end.

Notable

A member traces what happens to miner collateral alpha when a subnet is pruned, working through the mechanics step by step with a moderator.

Opening ·

A member asked when miner's collateral alpha is released as incentive, does it replace the incentive until all the collateral is unlocked? or is it released in addition to earning incentive?

Clarification ·

A moderator replied I don't know what you mean by "collateral alpha". This sounds like it might be a subnet-specific function of a particular subnet's IM., and the member clarified they meant neuron registration collateral, part of reg cost that is locked instead of burn, which the moderator answered with a documentation link: It's explained here: https://www.bittensor.com/docs/guides/mining/collateral (for a subnet that has enabled it)

Follow-up ·

The member then asked I found the answer, but I can't find where the collateral goes when the subnet is pruned, and the moderator quoted the documentation directly: If the subnet dissolves, the lock metadata is cleared after alpha is settled to coldkey free TAO with everyone else's stake — there is no separate collateral refund step., adding that "Dissolved" is another term for "deregistered", which is another term for "pruned".

2 things to read from it

  1. The exchange is a plain mechanism clarification rather than a policy debate, but it is notable that the moderator did not know the term 'collateral alpha' at first and had to be pointed to the specific subnet feature by the asker before locating the documented answer.
  2. The resolution confirms that pruned-subnet collateral has no separate refund path -- it is settled to coldkey free TAO alongside everyone else's stake -- a detail not previously surfaced in this channel.

Where it stands

Resolved: the member confirmed I see, thanks after the moderator supplied the documented answer.

Notable

A member credits a named figure with following through on a prior promise to bring the subnet registration fee down to around 100 TAO.

Claim ·

A member wrote, pointing to an on-chain event, that the figure previously said he would bring the registration cost down to around 100 TAO. Looks like he's followed through on that promise.

2 things to read from it

  1. This follows Nov 9's self-driven walk-through of the registration-fee ratchet mechanism, where a member identified the sudo-set parameter behind the decline but could not get an answer on who had authority to change it; today's claim names a person as having intended the drop, though the claim is not independently confirmed in the material.
  2. No moderator or other member in #General corroborated or disputed the attribution today, leaving the link between the earlier unanswered sudo question and today's claimed promise unresolved.

In the week’s context

  • Nov 9: a member traced a subnet registration-fee drop to a specific sudo-set parameter and described a one-way pricing ratchet, with a pointed question about who executed the change going unanswered.

Where it stands

Open: the claim that a named figure fulfilled a prior promise is unverified in the channel, and the underlying question of sudo authority from Nov 9 remains unaddressed.

Other

Other: a confused member's questions about buying subnet 78 draw a FAQ redirect and a scam report, the standing scam warning repeats, and routine greetings and a ban request round out the day.

Mixed ·

A member asking How can I check this tell me the process as I want to buy a subnet like 78 was pointed to the documentation: Please see the documentation I'd start with the "Subnet Development" section of the FAQ, which will lead you to the official documentation for information on developing and registering subnets. You can find the FAQ in <#1215386737661055056>, shortly after the same member asked moderators Please ban this scammer. Separately, a moderator repeated the channel's standing reminder: People claiming to be mods, admins, Opentensor Foundation members, subnet support, or any kind of "support" sending you a direct message or friend request -- they're imposters, block them!, and told a member celebrating a scam attempt Hah, congrats (?) -- make sure you block and report to Discord!

What to read from it

  1. Volume held near its weekly high at 24 messages, continuing from Nov 9's 30, with most of the remaining traffic being routine moderation, redirects and greetings rather than new ecosystem ground.

Where it stands

Closed through standard redirect, reminder or silence.

#Rao

2 REPORTS

Notable

The root-basket deposit freeze gets its first concrete fix estimate, offered by a community member rather than the team

Opening ·

A member picked up the running complaint directly, asking When will this be fixed? The balance in my basket still hasn't changed.

Reply ·

Another member answered with a specific pointer rather than a repeat of the standing line: its in the release branch. My guess is it will go out tomorrow., linking a GitHub pull request.

3 things to read from it

  1. This is the first time in the multi-day freeze thread that anyone in #Rao has pointed to an actual artifact -- a release-branch PR -- rather than restating that a fix is 'being worked on.'
  2. The estimate is still a guess from a community member, not a confirmed timeline from the team; #Rao carried none of the freeze discussion on Nov 9, so today's exchange is the channel's first movement on the issue in two days.
  3. The same day, #General logged nothing new on the freeze beyond a one-line 'I think rewards are back' comment, leaving #Rao as the only place where the fix's mechanics were discussed.

In the week’s context

  • Nov 7's #Rao thread first got the team to confirm deposits were blocked while dividends kept accruing, with no timeline given.
  • Nov 8's #Rao thread moved into a design critique of the blocking gate itself, again with no team response.
  • Nov 9 saw the freeze resurface only in #General, with a new symptom (a basket appearing to vanish) and no fix timeline from the moderator.

Where it stands

Open; the PR exists in the release branch per a community member's reading, but no official confirmation or firm release date appears in the material.

Other

Other: a member shared a Swarm buyback-and-burn announcement with no discussion

Share ·

A member posted a link to an SN124 (Swarm) update describing changes including 20 TAO buyback & burn and an Ongoing buyback & burn program.

What to read from it

  1. The link drew no reply in the channel, and #Rao's total volume stayed at its lowest point of the week.

Where it stands

Closed; a one-line share with no follow-up in the channel.

Subnets

5 REPORTS

SN124 24h +4.14%

Swarm

What happened
A miner raised concerns that the leading Swarm model may have known evaluation seeds in advance, pointing to large score swings between epochs for the top-ranked UID. A community report further alleged that one coldkey controls roughly 60% of SN124's Swarm Sentinel seats through multiple hotkeys and called for public Sentinel scoring. SN124 (Swarm) separately announced a 20 TAO buyback and burn, an ongoing buyback program, restriction of subnet output to paying clients only, and increased staker allocation, while discontinuing the SAR and Swarm SAR tasks.
Community
Miners questioned the score volatility of the top model and asked whether it had prior knowledge of validation seeds; one community member published a detailed report on Sentinel seat concentration and pushed for making Sentinel scoring public. Others asked why emissions were affected after the SAR and Swarm SAR tasks were stopped.
Official
The team said it reviewed whether seeds had leaked and concluded they had not, attributing the score swings to the top model overfitting to compute specifics; it declined to make Sentinel scoring fully public, citing a client confidentiality agreement and competitive concerns, while confirming the buyback, client-only output policy, and the end of SAR and Swarm SAR tasks.

SN44 24h −0.10%

Score

What happened
The SN44 (Score) team removed multiple evaluation blocks across the Room Dirtiness, Beverage, Vehicle and Petrol tracks after miners disputed ground-truth annotations. All public evaluation elements went offline for an extended period before the team confirmed the pipeline was fixed. The team also declined to remove a block on the fallen-person track despite a miner's zero-score dispute, citing the requirement to keep chutes in a hot state. Separately, two clusters of flagged solutions on the cricket private track were blacklisted again for rule violations.
Community
Miners repeatedly flagged suspected ground-truth errors across several tracks and asked the team to clarify or remove specific blocks. Some miners pressed for more transparent annotation rules, while others raised concerns about the outage affecting pipeline stability.
Official
The team removed a number of specific blocks across Room Dirtiness, Beverage, Vehicle and Petrol, stated other challenges were correctly annotated, and reiterated that chutes must remain in a hot state rather than removing contested blocks.

In debate · SN44 ground-truth annotation disputes · 14 participants

  • Miners argue certain ground-truth annotations are unclear or incorrect and want blocks removed or clarified.

    “I think it must be removed or you should share how to annotate this image with miners for transparency.”
  • The team maintains most flagged cases are edge cases that do not qualify for removal.

    “no 2 pixels merged with black on the very edge of an image that could be anything really”
  • Some miners push back, saying the team's explanations do not meet their stated transparency goals.

    “Miners just want to understand exactly why something is being rejected, but you keep saying it's an edge case, so it can't be removed.”

SN108 24h −2.16%

ChipForge

What happened
A miner's design scored 87.2, far above the previous best of 12.21, prompting questions from community members about the result's validity. The team said the score passed initial checks and placed it under a 24-hour probation review before confirming it was valid under current rules. ChipForge extended Quest 2 by 7 days to Nov 23, pausing and then reopening submissions, and later added a feature letting miners withdraw a submission from the queue before a validator picks it up.
Community
Several miners questioned whether the operations team could view submissions before validators did, since entries go directly to the team's own server, and asked for on-chain hashing or encryption as a safeguard. The miner behind the top-scoring design later stated publicly that he has no connection to the team.
Official
The team said its initial review found the 87.2 result valid under current rules and that it passed all checks, and explained that on-chain hashing would not resolve the trust question since it only proves a file existed at a given time.

In debate · ChipForge 87.2 score dispute · 10 participants

  • A miner argued that the operations team could see submissions early because entries go directly to its own server.

    “subs go straight to their own server (api.chipforge.io) and the owner runs a validator too, so they can always see them first”
  • The team said its initial review found the top score valid under the current rules and urged miners to keep participating.

    “Our initial review shows the 87.2 is a valid result under the current rules, and it passed all checks. Keep mining.”
  • The miner behind the top-scoring design denied any insider connection to the team.

    “I'm an independent miner with no connection to the team.”

SN61 24h −3.38%

RedTeam

What happened
A dashboard issue left BEX challenge scores calculated only for UIDs up to 190 for several hours; the team confirmed the cause and said scoring was later completed. Separately, the team announced a new BV-2 challenge, which will replace the ADA-3 challenge rather than BEX-1, with a target release within the week.
Community
Miners repeatedly asked the team to check the scoring issue, and one community member asked the team to clarify what had gone wrong and what the new challenge would involve. A moderator later intervened after repeated personal attacks and spam in the channel, warning of timeouts if the bickering continued.
Official
The team confirmed the dashboard scoring issue, said BV-2 will replace ADA-3 (not BEX-1), and targeted a release within the week.

SN36 24h +52.91%

Weir

What happened
SN36 (Weir) remains without a confirmed owner in its official channel, yet on-chain data shows active mining and emissions, reported by community members at around 20 TAO/day. Moderators confirmed the subnet owner has not claimed the channel or provided support there, and said on-chain identity fields shown on block explorers can be set arbitrarily by whoever holds the key, calling such sources low-confidence. No individual or team has stepped forward to confirm ownership or explain the emissions.
Community
Community members questioned how an unclaimed subnet could be earning emissions and asked moderators to raise the matter with the foundation, with one suggesting emissions be cut until ownership is clarified. Moderators repeated the channel rule that anyone claiming subnet ownership, or relaying such a claim, will be timed out.
Official
Moderators stated they have no information on how the emissions are occurring and said questions on cutting emissions would need to go to the foundation contact, noting they are not OTF or Rao Foundation employees.

More subnet community coverage every day in the Telegram channel →

Ecosystem debate

3 participants

Unsolicited subnet slot purchase offers across multiple channels

A single account posted identical messages offering to buy a subnet slot in at least ten different subnet channels, including SN101, SN47, SN84, SN26, SN109, SN57, SN40, SN87, SN113, and SN96. The repeated, templated wording across unrelated subnets drew attention from other community members and moderators.

Positions

  • The account repeatedly solicited subnet owners to sell their slots via direct message, using the same message in channel after channel.

    “Hi gm, I'm looking to buy a subnet slot. Would you be interested in selling this subnet? If so, please send me a DM.”
  • A community member suggested such outreach is indistinguishable from mass spam targeting inboxes.

    “I'm 100% sure that your DM box is flooded with spam accounts”
  • A moderator clarified the account has no affiliation with the subnet teams being approached, directing users to official pinned information instead.

    “This user is not a member of the subnet team. See the pinned posts for information about the subnet and its team.”
Official response
There has been no official response from Opentensor or the named subnet teams; the only intervention noted is a moderator flagging the account as unaffiliated with the subnet team.
What to watch
Whether the account continues posting in additional subnet channels or whether moderators take further action.

Voices on X

  • @opentensor

    Opentensor introduced Gamma Tokens, a payment system letting users and AI agents pay for subnet services directly from a Bittensor wallet without separate accounts or API keys.

  • @CoinAsia_io

    CoinAsia reported that Tencent selected SN64 (Chutes) as a launch partner for its Hy4 AI model, citing Chutes' inference pricing of $0.525 per million input tokens and $1.5625 per million output tokens.

  • @robotensor_ai

    Robotensor said SN58 (Attune) is the first subnet to enable Gamma, launching a live service that lets users pay with Gamma for robotics training data collected through Horizon.

  • @ShizzyUnchained

    A community member said the GammaEnabled flag switched from false to true network-wide at block 9,471,027, but that no subnet had recorded a Gamma opt-in as of a later snapshot.

  • @JesusMartinez

    A user said SN92 (REAX)'s earlier model ranked #2 on an unspecified decision-model ranking and that its newer REAX 1.3 model, trained via Bittensor model competitions, now ranks #1.

  • @near_ai

    Near AI said an attacker exploited a vulnerability in its SGLang inference engine on November 8 to access part of its infrastructure, adding that access has been closed and affected users contacted.

  • @taodaily_io

    The TAO Daily reported that SN75 (Hippius) launched a mobile app offering encrypted file storage, automatic photo backups and 10 GB of free storage.

  • @cascade_sn91

    SN91 (Cascade) said its new Cascade Mixture-of-Experts forecasting models reached an 18.4% gap to Timestral-209M on the GIFT-Eval CRPS benchmark and 19.1% on MASE after 10 days of competition.

◆2026-11-11 · End of this edition

Protocol

9 PRs

New release · v489 Not released on its own

Subtensor v489: Gamma emission funding, lease stake recovery, and root deposit queue fixes

This release lets Subnet Owners direct part of their Alpha Emission to prepaid Gamma service credits on other Subnets. It also returns leftover Stake to the beneficiary when a Subnet lease ends, and restores Root staking for hotkeys with oversized pending deposit queues. Gamma deposit creation now has ownership checks, and Root dividend admission uses new price-impact limits.

Read the upgrade report →

#3308 Breaking change Affects emission

Release 489

What
This release fixes the Python SDK so it correctly handles transactions signed with externally held ECDSA (secp256k1) keys, including those stored in cloud hardware security modules.
Why
The SDK was stripping a required byte from valid ECDSA signatures before submission, so transactions signed outside the SDK with this key type could not go through even though the chain itself already accepted them.
Who it affects
This affects developers and services that sign transactions with external ECDSA keys or HSMs rather than wallets stored by the SDK itself; ordinary TAO holders, validators and miners using standard wallets see no change.

#3311 Breaking change

Fix lease stake recovery and aggregate stake gas metering

What
This PR fixes an issue with recovering stake tied to subnet leases and corrects gas metering for an aggregate staking function exposed through the EVM, while also disabling an older stake-share compatibility mapping in production.
Why
Stake recovery for leased subnets and gas costs for the aggregate staking call were not working as intended, and the legacy mapping was kept only for compatibility reasons that no longer apply in production.
Who it affects
This affects subnet owners and lessees with leased subnets whose stake recovery depends on this mechanism, and developers calling the aggregate staking function via the EVM; it does not change normal staking or validating for most TAO holders.

#3313

fix(staking): recover oversized pending basket deposit queues

What
This PR fixes a bug where a hotkey holding more than 256 pending basket deposit entries would get its root-stake operations permanently stuck, with the background process unable to clear the backlog.
Why
The system had a fixed limit on pending deposit entries per hotkey, and exceeding it left the queue in a state the automatic drain process could not resolve.
Who it affects
This affects hotkeys that had accumulated an unusually large number of pending basket deposits and were unable to complete root-stake actions; most validators and miners with normal deposit activity are unaffected.

#3314

fix(gamma): restore independent API-key deposit ownership

What
This PR removes a check that required a hotkey used for Gamma deposits to already have a staking relationship with the funding coldkey before opening a deposit.
Why
The check contradicted the intended design, under which an API-key hotkey does not need to be staking-linked to the coldkey funding it.
Who it affects
This affects users and services opening Gamma deposits with separate API-key hotkeys; it restores a previously intended independence between the funding coldkey and the deposit hotkey.

#3320 Breaking change Affects emission

Add gamma emission: subnet emission shares that buy other subnets' gamma

What
This PR adds a gamma emission mechanism letting a subnet owner direct a share of their subnet's alpha emission to buy and burn another subnet's alpha, crediting the resulting gamma to the owner's deposit on that target subnet, with root controlling which subnets can receive it.
Why
The feature lets a subnet pay for services it consumes from another subnet directly out of its own emission, each share automatically buying and burning the target subnet's alpha.
Who it affects
This affects subnet owners who set emission splits and the subnets they target for gamma; validators and miners on affected subnets may see changes to how emission is split since gamma shares reduce the amount going to the fixed owner/validator/miner/burn/recycle slots, while root whitelisting determines which subnets can participate at all.

#3312

Sdk pure key helpers

What
This PR adds helper functions to the SDK for strict signature verification without a signature-wrapping fallback, and for encoding or decoding passwords without touching process environment variables.
Why
These helpers give developers safer, more predictable primitives for signature checking and password handling within the SDK.
Who it affects
This affects developers building on the SDK; it introduces no change for ordinary TAO holders, validators or miners interacting with the network directly.

#3322

docs: extend the services document and link it to signed requests

What
This PR extends the services documentation format with optional fields covering published service listings, maximum per-request spend for gamma payments, a named signing protocol for gamma requests, and service name and description fields, and links it to the signed-requests guide.
Why
The added fields give gamma-based services a documented way to limit spend and sign requests, while remaining backward compatible since all new fields are optional.
Who it affects
This affects subnet teams and services that publish a services document for gamma payments; TAO holders, validators and miners are not directly affected beyond relying on services that may now document themselves more completely.

#3324

hotfix(sdk): bump bittensor to 11.3.3.dev0

What
This is a hotfix that bumps the SDK's package version number after the prior release's version had already been published, which had blocked the release pipeline.
Why
The release pipeline failed because the SDK version in the project files had not been incremented past a version already published on PyPI, which included the new ECDSA, password-helper, and gamma-related changes from the prior release.
Who it affects
This affects only the release process and SDK packaging; it has no functional effect for TAO holders, validators or miners.

#3325 Breaking change

Release 490

What
This release increments the runtime's spec version to 490 so the release tag correctly reflects the updated SDK version, without changing any runtime logic from the previous release.
Why
The prior spec version's pending proposal will not be signed, so the version number needed to be moved forward to keep the release tag consistent with the SDK bump.
Who it affects
This is a release-bookkeeping change only; it affects no one outside the development and release process, since the actual runtime behavior is unchanged from the previous release.

Recent editions

Past editions +