Daily digest of the Bittensor ecosystem

中文/EN

Daily edition
Nov 13, 2026

Bittensor Daily A 1T1B product
1T1B ↗
BITTENSOR DAILY

24H Market

01/10

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

THE DAILY BRIEFING

Bittensor Daily — November 13, 2026

#Rao's quantum-resistant wallet thread finally moved from promise to product, walking through key-type tradeoffs in detail, even as #General logged a pointed complaint about a practice the channel would not engage with substantively. Subnet news carried more weight than the ecosystem channels: Lium published strong 30-day growth and buyback figures, while TPN Labs and Fugal each faced unresolved questions about originality and hotkey concentration. The root-basket freeze that ran all last week did not resurface in either channel today.

Today's highlights

  1. The TAOApp wallet extension goes live with quantum-resistant key support, and a #Rao member walks through which key type to actually choose.

    It's live, you can use quantum-resistant wallets (and all the other key types) in the TAOApp wallet extension now 🫡 and, on which to pick, I would personally choose `ML-DSA-65`. P.S Someone who trades a lot, and wants lower transaction fees might prefer `Falcon-512` (cheapest of the new types) or `Falcon-1024`. (1) This moves the rollout from announcement to usable tooling: on Nov 7 the same channel could only say the team was working on extension support, and on Nov 6 a member flagged hardware-wallet lag as the real bottleneck. (2) The answer confirms existing hotkeys or subnets migrate via a coldkey swap rather than in-place conversion -- detail not previously surfaced. (3) The hardware-wallet and Polkadot.js gaps flagged earlier in the week were not revisited today; the asker thanked the member and closed the exchange.

  2. A member calls self-mining a systemic threat driving miners out of Bittensor, and is told only to stop tagging admins.

    the practice of "self-mining"—carried out under the guise of a "winner-takes-all" dynamic—is dealing a severe blow to the entire Bittensor ecosystem and Many of the miners who have long been part of Bittensor are falling victim to this practice and choosing to leave. A moderator replied Hi, you don't need to tag admins to this. You can air your opinion without tagging admins., and a later asker's Can anybody let me know how I can report the sign of self-mining was answered only with FAQ links. (1) Two separate members raised the same concern hours apart without any connection drawn between them, suggesting it is a live worry rather than one complaint. (2) No official voice confirmed or addressed whether such a practice occurs on any subnet. (3) Both exchanges closed on etiquette or documentation rather than the underlying claim.

  3. SN51 (Lium) reports a 30-day surge in paying customers, deposits and run rate alongside a roughly 5,000 TAO buy-and-burn, while also cutting its own owner emission share.

    The team stated new paying customers up 78%, deposits up 94%, existing customer spend up 46%, annualized run rate up 79%, GPU spend up 87% and rentals up 67%, plus a buy-and-burn of about 5,000 TAO on a $1.3 million cost basis. (1) This follows the week's recurring community pressure on Lium over whether rental fees and emissions can ever net positive for the token, including Nov 9's question about an unpublished buyback plan. (2) The owner share fell from 18% to 9% while miners rose from 41% to 50%, validators unchanged at 41%. (3) The subnet's token rose 7.18% over 24 hours; no further community reaction to the new figures appears in the material.

  4. On SN65 (True Performance Network), a miner presents a per-tensor hash comparison alleging the top two ranked TPN-008 submissions are near-identical to their own earlier model.

    (1) The miner pointed to repositories created close in time to each other as evidence, while other miners separately raised concerns about models being forked after the evaluation window closed. (2) This follows TPN-008's public release on Hugging Face, where it scored 94.36% on a sealed verified benchmark while compressed to under 25.1MiB. (3) The team acknowledged the concerns and said it needs time to reach internal consensus before deciding on a course of action; the disputed payout was continuing to run as of the discussion, and the token rose 3.73% over 24 hours.

  5. On SN70 (Fugal), one operator's control of 51 hotkeys draws community scrutiny, while a separate validator timeout bug is confirmed and patched.

    (1) Community members questioned the concentration after observing the operator runs 9 active hotkeys out of 51, with the remainder described as idle v1 hotkeys. (2) The operator said each active hotkey runs a different head with its own hash and that standby hotkeys are counted separately by the protocol, while the team confirmed a validator timeout bug causing some miners to miss about 25% of epochs, fixed in v0.3.2. (3) The team said the current leader's position would only reset with a base model change, not the upcoming v2.1 update; the token fell 6.26% over 24 hours.

Discord conversations worth following

#General

2 REPORTS

Notable

A member calls 'self-mining' a systemic threat to the ecosystem, and is told to raise it without tagging moderators

Opening ·

A member argued that the practice of "self-mining"—carried out under the guise of a "winner-takes-all" dynamic—is dealing a severe blow to the entire Bittensor ecosystem, adding that Many of the miners who have long been part of Bittensor are falling victim to this practice and choosing to leave and that the value of the entire Bittensor ecosystem is simply the sum of the quality of its individual subnets.

Reply ·

A moderator did not dispute the claim but redirected the manner of raising it: Hi, you don't need to tag admins to this. You can air your opinion without tagging admins. Please review the.

Follow-up ·

Later the same day, another member asked Can anybody let me know how I can report the sign of self-mining, and was pointed to two FAQ posts rather than given a direct answer in-channel.

3 things to read from it

  1. Two separate members raised the same concern hours apart without any connection between them being drawn in the channel, suggesting it is a live worry among users rather than a single complaint.
  2. Per the publication's own rules, the substance of the accusation is reported only in the member's own words; no official voice confirmed or addressed whether such a practice is occurring on any specific subnet.
  3. The moderator response addressed channel etiquette (tagging) rather than the underlying concern, and the reporting request was answered with a document pointer rather than a described process.

Where it stands

Unresolved as a substantive question: the channel directed both members to existing FAQ material and etiquette rules, but no official reply engaged with the claim that this practice is driving miners to leave.

Other

Other: routine support, scam warnings and greetings filled most of a quiet #General day

Support ·

Members asked how to join the dTAO community, how to get testnet TAO, and whether the Talisman Discord was working, each answered with a redirect — one asker posted a wallet address hoping someone is so kind of sending a bit, which the channel did not fulfill and instead pointed to the FAQ.

Moderation ·

The standing scam warning ran twice more today, including the line if you get any DMs or FRs from people claiming to be support for Talisman (or any kind of support at all, including people pretending to be me or other mods), it's a scam.

Other ·

A member asked Are there any bittensor investor on nvda board btw and was told the server does not handle investment discussion; a newcomer greeting and a member's praise for a community-made song — calling it a beautiful expression of love for the TAO community— rounded out the day.

What to read from it

  1. Volume held at 24 messages, matching Nov 10's level and well below Nov 11's 29 and Nov 9's 30, with nothing in today's remainder touching tokenomics, governance or protocol direction.

Where it stands

All items closed through standard redirect, reminder or the standing scam warning; nothing substantive remains open.

#Rao

2 REPORTS

Notable

Quantum-resistant wallets go live in the TAOApp extension, and a #Rao exchange works through which key type to actually pick

Opening ·

A member announced the rollout had reached the wallet tooling itself: It's live, you can use quantum-resistant wallets (and all the other key types) in the TAOApp wallet extension now 🫡

Clarification ·

Asked whether an existing wallet could be converted, the same member explained You can create a new quantum-resistant wallet in the extension, and transfer things over there (and if one have hotkeys and/or a subnet etc. then one would just do a coldkey swap into a new quantum-resistant wallet), adding that the extension lets a user pick between ML-DSA-65, Falcon-512 or Falcon-1024.

Reply ·

Pressed on which type to use -- Which one should one use? Can't be all? -- the member suggested that for maximum security you could do a multi-sig, where the different signatories would use different key types, and gave a personal pick: I would personally choose `ML-DSA-65`. P.S Someone who trades a lot, and wants lower transaction fees might prefer `Falcon-512` (cheapest of the new types) or `Falcon-1024`.

Outcome ·

The asker thanked the member by name and closed the exchange: Thanks Maciej.

3 things to read from it

  1. This moves the quantum-resistant wallet rollout from announcement to usable tooling: on Nov 7, the same channel could only say the team was 'already working on the support for them in the TAOApp Extension'; today that extension is live and a member is walking through its actual key-type tradeoffs.
  2. The answer fills in practical detail none of the official material has offered so far -- that a coldkey swap, not an in-place conversion, is how existing hotkeys or subnets migrate to a quantum-resistant coldkey.
  3. The thread does not touch the two gaps flagged earlier this week -- hardware wallet support (Nov 6) and third-party tools like Polkadot.js (Nov 7) -- so those remain open even as the native extension path is now confirmed working.

In the week’s context

  • Nov 6's #Rao thread first detailed that 'hardware wallet lag is the real bottleneck' and that root had not yet enabled hashed accounts for real transactions.
  • Nov 7's #Rao exchange on Polkadot.js support got only a promise that the team was 'already working on the support for them in the TAOApp Extension' -- today's thread shows that extension now live.

Where it stands

Resolved for the native-extension path: wallet creation and key-type selection are live and explained. The hardware-wallet and Polkadot.js questions raised earlier in the week were not revisited today.

Other

Other: a member flagged the SN3 owner-key compromise report with a one-word reaction, and Novelty Search's episode was cancelled for the night

Share ·

A member posted the TensiaFDN thread on the SN3 owner-key incident with no further comment beyond another member's response, Very concerning.

What to read from it

  1. The reaction is brief and adds nothing beyond what was already reported on Nov 11, when the owner confirmed the breach and an independent investigation traced the stolen alpha.

In the week’s context

  • Follows Nov 11's confirmed SN3 (Teutonic) owner-coldkey compromise, in which 101,141 alpha was moved and sold for roughly 2,068 TAO.

Where it stands

Closed; no further discussion followed in the channel.

#Announcements

1 REPORT

Other

The only #Announcements message today was a one-line cancellation notice for Novelty Search

Notice ·

A member posted that Novelty Search has been cancelled for tonight, catch up on previous episodes via Opentensor YouTube.

2 things to read from it

  1. The channel produced a single routine cancellation notice, continuing the near-silent run #Announcements has shown through this stretch of the week; nothing tokenomics-, governance- or mechanism-related surfaced here today.
  2. The cancelled event is the same Novelty Search series that #General and X promoted on Nov 5, but today's note is purely administrative and carries no new content about the series itself.

Where it stands

Closed; a cancellation with no further detail and no replies in the channel.

Subnets

5 REPORTS

SN42 24h +16.22%

CADmind

What happened
SN42 (CADmind) launched on Bittensor mainnet and joined the Yuma Accelerator program, with miner emissions enabled within hours of securing its subnet slot. The team onboarded an initial whitelist of 5 miners during a 2-week calibration phase, running through December 1, after miners passed an onboarding test. A hyperparameter change on SN42 reduced max_validators from 64 to 1.
Community
Some miners said they had already paid registration fees and set up infrastructure before learning that onboarding would be limited to only 5 slots, and argued this should have been communicated earlier. The team described one miner's response in the channel as inappropriate.
Official
The team acknowledged that not all communications went out before miner submissions began, said building a quality service is its primary concern, and stated more miners will be accepted as the userbase grows.

SN70 24h −6.26%

Fugal

What happened
SN70 (Fugal) community members raised questions after observing that one operator controls 51 hotkeys, of which 9 are currently active, with the remaining 42 described by the operator as idle, mostly old v1 hotkeys without a server behind them. The team separately identified a validator timeout bug causing some miners positioned later in UID order to miss about 25% of epochs even when proofs were ready on time, and released a fix in v0.3.2. The team said the current leader's position would only reset with a base model change, not with the upcoming v2.1 update.
Community
Several miners said they see little room to improve on the current leader's results and questioned why one operator holds the majority of hotkeys. A validator called for a fairer reward structure, citing concerns from past experience on other subnets, and said they would not resume work until changes are made.
Official
The operator holding the 51 hotkeys said only 9 are active, each running a different head with its own hash, and said standby hotkeys are counted separately by the protocol. The team confirmed the validator timeout bug, released a fix in v0.3.2, and urged validators to upgrade.

SN51 24h +7.18%

lium.io

What happened
SN51 (Lium) published a 30-day update showing new paying customers up 78%, deposits up 94%, existing customer spend up 46%, annualized run rate up 79%, GPU spend up 87% and rentals up 67%. The team also bought and burned approximately 5,000 TAO on a cost basis of about $1.3 million, and said further buy-and-burn activity would follow using idle GPU mining revenue and other revenue. Separately, SN51 changed its emission split, reducing the owner share from 18% to 9%, keeping validators at 41%, and raising miners from 41% to 50%.
Community
Several community members had asked about activity in the owner wallet, including questions about a prior token sale and whether the coldkey had been compromised. After the buy-and-burn announcement, reactions were mixed: some praised the move and the monthly update, while others said the earlier sale had already led them to sell holdings and asked for clearer communication going forward.
Official
The team stated it had bought and burned about 5,000 TAO on a $1.3 million cost basis and plans further buy-and-burn using idle GPU mining revenue and other revenue.

In debate · Owner wallet activity and the buy-and-burn response · 9 participants

  • Some community members questioned unexplained owner wallet activity, including asking if the coldkey was compromised.

    “coldkey hacked?”
  • A community member said the lack of earlier explanation had already caused them to sell their holdings.

    “Thank you but you didn't cleared why you sold 5k tao earlier because of that we have to sold our holdings”
  • Others welcomed the buy-and-burn as a positive sign and praised the team's delivery on community requests.

    “Verified the buy and burn on chain: ~5k τ in, ~43k α gone for good. That's real conviction.”

SN65 24h +3.73%

True Performance Network

What happened
TPN Labs' 8th model (TPN-008), based on Google Gemma 4-31B, went live on Hugging Face, compressed to under 25.1MiB while scoring 94.36% on a sealed verified benchmark, with the repository now public and open-weight. Following the competition, a miner presented a per-tensor hash comparison arguing that the top two ranked TPN-008 submissions were near-identical to their own earlier model, noting the repositories in question were created close in time to each other. Other miners separately raised concerns about models being forked after the evaluation window closed.
Community
Several miners and community members reacted to the evidence, questioning the fairness of the competition and calling for action against the submissions in question. Some also asked how emissions and benchmark credit were being distributed given the dispute.
Official
The SN65 (True Performance Network) team acknowledged the concerns raised and said it needs time to reach internal consensus before deciding on a course of action; the disputed TPN-008 payout was continuing to run as of the discussion.

SN120 24h −1.52%

Affine

What happened
SN120 (Affine)'s evaluation pipeline was repeatedly stuck for hours and, at times, days, leaving hundreds of UIDs unevaluated and blocking newly registered miners from being scored. Miners also disputed whether a dominant participant was controlling many UIDs through bulk registrations and submissions. A miner argued that pass-rate and benchmark scores had plateaued over roughly 200 epochs and proposed a breaking change to the mechanism, which the team discussed alongside eval-speed fixes. Separately, speculation circulated that the registration fee would rise to 1.5 TAO amid an active registration war.
Community
Miners repeatedly pressed the team to fix the eval stalls, with some suggesting the pending submission be skipped to keep the queue moving and others calling for emissions to be burned to curb bulk registration. Several participants accused a dominant UID holder of monopolizing slots through bulk submissions, while others pushed back, saying miners should wait their turn rather than complain.
Official
The team acknowledged the eval system is too slow to capture recent training improvements and said it has already increased the eval dataset size and adjusted thresholds to reduce noise.

In debate · SN120 (Affine) eval stalls and bulk-registration fairness · 9 participants

  • Miners said the slow, stalled evaluation queue left hundreds of UIDs unprocessed and gave new registrants no chance to be scored.

    “Could you speed up the evaluation? There are nearly 300 non-evaluated UIDs, but only 15 were evaluated in the last 24 hours.”
  • The team said performance gains exist but the current eval system is too slow to measure them, and cited dataset and threshold changes already made.

    “I think we're seeing good performance improvements in affine training. but our current eval system is too slow to capture them properly.”
  • Some community members argued miners should simply wait in queue order rather than demand submissions be skipped.

    “why skip? wait in order, not complaining.”

More subnet community coverage every day in the Telegram channel →

Voices on X

  • @TensiaFDN

    @TensiaFDN says the keyholder behind SN46 (Instant) received about 720,000 alpha over fourteen months and moved 808,918 alpha in 58 transfers to 17 wallets, with proceeds of about 2,750 TAO landing in one Coinbase account.

  • @byteleap_ai

    @byteleap_ai reports SN3's owner key was compromised, with 101,141 alpha moved to a new wallet and sold within minutes for about 2,068 TAO, pointing to loose key management as the likelier cause.

  • @lium_io

    @lium_io's 30-day update shows new paying customers up 78%, deposits up 94%, existing customer spend up 46%, annualized run rate up 79%, GPU spend up 87% and rentals up 67%.

  • @opentensor

    @opentensor highlights SN51 (Lium) generating nearly $35M in annualized GPU rental spending from a single Bittensor subnet.

  • @CAD_mind_subnet

    @CAD_mind_subnet announces CADmind is live on Bittensor, letting users create custom physical designs like parts, cases, drones and robotics from prompts, backed by @YumaGroup.

  • @TPN_Labs

    @TPN_Labs says Competition 8 has ended, with TPN's 8th model, based on Gemma 4-31B, now public on Hugging Face at under 25,1MiB and scoring 94,36% on a sealed verified benchmark.

  • @cathedralhq

    @cathedralhq ran 1,000 AI agents in isolated, network-off TEE sandboxes with signed records for each run, noting some solved their task, more failed, and one run's seal could not be proven.

  • @TensiaBot

    @TensiaBot reports SN51 (Lium) changed its emission split, with the owner share falling from 18% to 9%, validators unchanged at 41%, and the miner share rising from 41% to 50%.

◆2026-11-13 · End of this edition

Recent editions

Past editions +