Daily digest of the Bittensor ecosystem

中文/EN

Daily edition
Nov 9, 2026

Bittensor Daily A 1T1B product
1T1B ↗
BITTENSOR DAILY

24H Market

01/10

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

THE DAILY BRIEFING

Bittensor Daily — November 9, 2026

The root-basket deposit freeze dominated both ecosystem channels for a second straight day: #General stayed at the symptom level with a confused user and a repeated known-issue line, while #Rao moved into an actual design critique of the price gate, unanswered by any team voice. Subnet news added a lium.io treasury-transparency question and an Affine gate that let a degenerate model pass, while the X feed carried Opentensor's weekly roundup crediting Reax, OpenRoboto, Macrocosmos and others. No new tokenomics or governance debate emerged beyond the basket freeze.

Today's highlights

  1. A #Rao member argues the Root-basket deposit gate is doing more harm than the bug it was built to stop, extending Nov 7's confirmed freeze into a second unresolved day.

    imo that freeze is doing more harm than the minting error its trying to stop one sn outside the band rarely moves nav enough to be worth blocking the whole flush pending piling up is the right call, i just wish more of it was landing (1) This is the first time a community member has challenged the gate's design rather than just asking when it will be fixed, following the team's Nov 7 confirmation that Root dividends are accruing but blocked from entering stakers' baskets. (2) A separate member described the mechanism itself, The deposits are still accumulating. There's just a funky gate blocking them from entering the basket, linking a technical write-up. (3) No team or official voice responded to the critique in the channel today.

  2. In #General, the same basket freeze leaves a user confused enough to accuse the channel of having no staking information, even after being pointed to the standing explanation.

    Anybody why my basket in roots not growing it stay in and not getting rewards? You guys here don't have no info about staking ? A moderator repeated: This is a known issue. Your basket is growing but you do not see it. It is an effect from the most recent upgrade. A fix is being worked on. (1) This mirrors Nov 7's independent #General report of an unclaimed basket stuck at zero, now continuing into a second day with no new timeline. (2) The gap between channels persists: #General offers only the generic known-issue line while #Rao is where the mechanism and its tradeoffs get discussed. (3) The user thanked the moderator and asked Thank you so is like a bug ?, closing the exchange without further detail.

  3. On SN51 (lium.io), a community member presses the team on whether buybacks and burns can ever return more TAO value than the subnet receives in emissions.

    A member cited about 617.07 TAO per day in incentive against roughly 116k USD in October fee revenue from 2.32M USD of renter spend, questioning the net-positive case for the subnet's token. (1) The team's channel account confirmed the 5% fee on October's renter spend but said no buyback plan or TAO-for-TAO comparison has been published. (2) Other members raised further questions on TAO billings held on the owner key, a formal treasury policy, and monthly reporting. (3) The team said these questions have been passed to the team, with a reply to follow once it responds.

  4. On SN120 (Affine), a GREEN-promoted checkpoint is found to consist entirely of one repeated token template, despite passing the admission gate.

    (1) The head model advanced through several checkpoints over the day, moving from opt 183 to opt 186, with most candidate rollouts scored RED and discarded while a handful of GREEN candidates were promoted. (2) A miner reported that a separate GREEN-promoted checkpoint, E953, consisted entirely of one repeated roughly 63-token template rather than genuine model generations, yet still passed the admission gate. (3) A community member said the issue is being looked into, with no fix confirmed in the material.

  5. Opentensor's weekly roundup credits Reax, OpenRoboto, Macrocosmos, Reliquary, Chutes, Trishool, ORO and Minos with the week's biggest stories, while X carried subnet buy/sell data and a Desearch product writeup.

    Reax produced the #1 AI decision model on Hugging Face's Decision Index, beating over 270 competing models. (1) The same account also highlighted OpenRoboto's commercial pilot with Vulcan Shield Hungary, Macrocosmos' IOTA Decide 4B model, and Reliquary's Teutonic-I-Reliquary result. (2) @taodaily_io separately reported SayGM, Vanta, Lium, ms.computer and Teutonic leading 24-hour net buys, with Score, Engy, Hippius, Minos and Affine leading net sells, and detailed Desearch's new Open Complaint Radar tool. (3) None of these figures were independently verified in the material.

Discord conversations worth following

#General

2 REPORTS

Notable

The root-basket deposit freeze flagged on Nov 7 resurfaces in both channels, now with a community member questioning whether the fix is worse than the problem

Opening ·

In #General, a member asked Anybody why my basket in roots not growing it stay in and not getting rewards? and, after being pointed to a pinned explanation, pushed back in frustration: You guys here don't have no info about staking ?

Clarification ·

A moderator repeated the standing answer: This is a known issue. Your basket is growing but you do not see it. It is an effect from the most recent upgrade. A fix is being worked on.

Expansion ·

In #Rao, a member described the mechanism behind the freeze: The deposits are still accumulating. There's just a funky gate blocking them from entering the basket, linking to a page titled root-dividend-flush-price-gate.

Pushback ·

Another #Rao member argued the fix itself is the bigger problem: that freeze is doing more harm than the minting error its trying to stop and one sn outside the band rarely moves nav enough to be worth blocking the whole flush, adding pending piling up is the right call, i just wish more of it was landing.

3 things to read from it

  1. This is the same deposit-blocking issue both channels independently described on Nov 7, now carrying into a second day with no fix shipped and a questioner in #General still confused enough to call the staking information inadequate.
  2. The #Rao exchange is the first time a community member has directly challenged the design of the blocking gate itself, rather than just asking when it will be fixed -- arguing the freeze causes more harm than the minting error it guards against.
  3. The gap between channels persists: #General still offers only the generic 'known issue, fix being worked on' line, while #Rao is where the mechanism (a NAV-band gate) and its tradeoffs are actually discussed.

In the week’s context

  • On Nov 7, a #Rao member asked why Root earnings weren't reflected in their basket and the team confirmed deposits were blocked pending a fix while dividends kept accruing, and the same day a #General member reported an identical symptom with their unclaimed basket stuck at zero.
  • This follows the week's run of breaking staking and basket changes -- Release 487's bundle and Release 488's partial reversal on Nov 5 -- though nothing in today's messages ties the gate explicitly to either release.

Where it stands

Still open: the moderator's answer in #General remains the same as Nov 7 with no new timeline, and in #Rao the design critique of the gate was not addressed by the team or anyone claiming authority over it.

Other

Other: a subnet-pick question, a moderator-recruitment ask, and a greeting rounded out a quiet day

Other ·

A member asked the channel What subnets you guys invested in? I like reax and open roboto without drawing further discussion; another asked How can I be moderator? and was told plainly There aren't any openings for moderators, sorry.; the day closed with a member wishing the channel Good day everyone, have a wonderful and productive week ahead 🫡

What to read from it

  1. Volume held at 15 messages, continuing the channel's run of support-and-routine traffic rather than any new ecosystem debate.

Where it stands

All items closed through brief replies or no reply; nothing substantive remains open.

#Rao

1 REPORT

Notable

Community members debate whether the blocked Root-dividend deposit gate is doing more harm than the bug it was meant to stop

Opening ·

A member described the mechanism behind the deposit block first confirmed by the team on Nov 7, saying The deposits are still accumulating. There's just a funky gate blocking them from entering the basket and linking a technical write-up of the gate.

Pushback ·

Another member argued the fix is worse than the problem: imo that freeze is doing more harm than the minting error its trying to stop, adding one sn outside the band rarely moves nav enough to be worth blocking the whole flush.

Concession ·

The same member still endorsed the broader approach of letting deposits queue rather than risk a bad mint, saying pending piling up is the right call, i just wish more of it was landing.

3 things to read from it

  1. This extends Nov 7's confirmation that Root dividends are accruing but blocked from entering stakers' baskets, now with community members debating the engineering trade-off behind the gate rather than just reporting the symptom.
  2. The same underlying deposit problem surfaced independently in #General today, where a member asked why their basket wasn't growing and was told it is a known issue tied to the recent upgrade -- the two channels again describe the same friction without cross-referencing each other, as happened on Nov 7.
  3. No team or official voice weighed in on today's #Rao exchange; the critique that the gate blocks the entire flush over a single out-of-band subnet came entirely from community members.

In the week’s context

  • On Nov 7, a #Rao member asked why Root earnings weren't reflected in their basket and the team confirmed deposits were blocked by a change it was fixing, while dividends kept accruing.
  • Nov 7 also showed the same issue reported independently in #General as a basket-display problem, a pattern that repeats today with a near-identical #General thread running the same day.

Where it stands

Still unresolved: the team has not responded to today's critique of the gate's design, and no timeline for lifting the deposit block or crediting the backlog has appeared in either channel.

Subnets

5 REPORTS

SN58 24h −1.48%

Attune

What happened
Attune (SN58) launched Horizon, a tool for teaching robots using phone-captured data. The team reported that its Vector Level 3 competition, running since Oct 29, saw the top score rise from 41. 7% to over 85%, with the top 4 competitors splitting 30% of miner emissions. Miners in the channel discussed moving on to a Level 4 competition, saying Level 3 is becoming difficult to improve on further.
Community
Miners disputed whether a top-ranked Level 3 submission was a copied model, with the accused miner denying the claim and asking for evidence.
Official
Robotensor AI (@robotensor_ai) announced the Horizon feature and reported Vector Level 3 progress, stating the crown score rose from 41. 7% to over 85% since the Oct 29 launch, with the top 4 champions sharing 30% of miner emissions.

SN31 24h −1.85%

rec4ll

What happened
SN31 (rec4ll) changed its min_burn hyperparameter from 500000000 to 250000000, a move intended to lower the registration fee. Community members disputed whether this actually reduced effective costs, pointing to the decay setting as a factor that kept fees elevated. A community member also noted that several coldkeys described as independent beta testers appear to have received their initial funding from the same source, questioning whether the project is a genuine product or an owner-run experiment.
Community
Reactions were split: some reported that registrations increased after the fee change, while others argued the effective cost had not dropped and raised questions about the legitimacy of the beta-tester setup.
Official
The team said the product is already working and is being rolled out with a free tier, and described the disputed test miners as recycling alpha each epoch for stability testing ahead of a final release.

In debate · SN31 (rec4ll) registration fee change and beta-tester legitimacy · 4 participants

  • A community member argued the fee cut was undermined because the decay setting was not adjusted, keeping effective costs high.

    “Cut in half but in the wrong way, you forgot to change the decay to be higher”
  • A team member said the fee change led to new registrations and defended the project as an active, functioning product.

    “The product is already working, we are starting with a free-tier to see how well it will scale on bittensor.”
  • A community member questioned whether the listed beta testers were independent, noting shared funding origins across coldkeys.

    “Whats interesting is that all of those coldkeys, have recieved first funds from the same source which means 1 person is running it all”

SN120 24h +0.33%

Affine

What happened
The AFFINE head model advanced through several checkpoints over the day, moving from opt 183 to opt 186 as most candidate rollouts were scored RED and discarded while a handful of GREEN candidates were promoted (including e903 at opt 185 and e934 at opt 186). A miner later reported that a separate GREEN-promoted checkpoint, E953, consisted entirely of one repeated ~63-token template rather than genuine model generations, yet still passed the admission gate.
Community
Several miners raised concerns that splicing or recycling rows from already-validated GREEN submissions let participants claim payee slots without producing new rollouts, while others defended the practice as a legitimate reuse of validated data. A miner also flagged the templated, non-model rows in E953 as worth investigating, without making further claims.
Official
A community member (u157) said the issue is being looked into.

In debate · Row recycling and payee-slot gaming on SN120 (Affine) · 8 participants

  • A miner argued that splicing rows from prior GREEN submissions lets participants grab payee slots without moving the model or incurring new cost.

    “recycling green rows is now an easy way to grab a payee slot.”
  • Another miner said the practice disadvantages those who actually rent GPUs and pay for fresh rollouts.

    “it's not that good for us miners who are actually renting GPUs and paying for the rollouts.”
  • A third miner countered that reusing already-validated data is comparable to a second training epoch and should not be treated as a problem.

    “If reusing data helps improve the model, I don't see why it should be a problem.”

SN44 24h −0.93%

Score

What happened
Miners on SN44 (Score) reported a series of ground-truth annotation errors across the Room Dirtiness, Petrol, Beverage and Vehicle tracks, flagging specific blocks they believed were mislabeled or produced incorrect false-positive results. The team reviewed the reports in batches and confirmed several blocks as incorrectly annotated, removing them from scoring, while stating other reported blocks were correctly annotated. Separately, miners reported the petrol track challenge element was not generating challenges, including one report that it was down for three hours.
Community
Several miners repeatedly pinged the team with block numbers and screenshots asking for re-checks, and some exchanges between miners over scoring interpretations grew tense before one participant asked others to stop arguing. The team pushed back on repeated pings, noting checks are performed twice daily and warning that continued excessive pinging could lead to unspecified measures.
Official
The team confirmed and removed several misannotated blocks in Room Dirtiness, Petrol, Beverage and Vehicle tracks across two separate review rounds, and asked the community not to double-ping on issues already checked.

SN51 24h −0.97%

lium.io

What happened
In the SN51 (lium.io) community channel, a community member pressed the team's channel account on how the subnet could ever return more TAO value through buybacks and burns than it receives in network emissions, citing on-chain figures of about 617.07 TAO per day in incentive versus roughly 116k USD in October fee revenue from 2.32M USD of renter spend. The discussion covered how providers are paid -- 95% of rental fees plus a share of a rented pool and an idle pool in the network incentive -- and whether any of that adds up to a net-positive outcome for the subnet. Other community members added questions about TAO billings held on the owner key, a formal treasury policy, and monthly reporting.
Community
Multiple community members pushed for a clear, published answer on how the subnet closes the gap between emissions received and value returned, saying the mechanics of renting, idle pay and bonus pools make this hard to assess from outside. Some framed a formal treasury policy and a stated path to net-positive status as something that would make the subnet easier for institutional investors to evaluate.
Official
The team's channel account said no buyback plan or TAO-for-TAO comparison has been published, confirmed the 5% fee on October's 2.32M USD of renter spend, and said the questions on treasury policy and a net-positive strategy have been passed to the team, with a reply to follow once the team responds.

More subnet community coverage every day in the Telegram channel →

Ecosystem debate

4 participants

Root basket deposit freeze / price gate

Following a recent upgrade, users reported that staking deposits into the root basket were not reflecting rewards. The team confirmed this is a known issue tied to a price gate that blocks the flush of accumulated dividends, with a fix in progress.

Positions

  • A user questioned why staked deposits were not receiving rewards and found no answer in the FAQ.

    “I told you my staking about 2 days are not paying I go to faqs no info too”
  • A community member explained that deposits are still accumulating but are blocked by a price gate from entering the basket.

    “There's just a funky gate blocking them from entering the basket”
  • A community member argued the freeze causes more harm than the minting error it is meant to prevent, since a single subnet rarely moves NAV enough to justify blocking the entire flush.

    “one sn outside the band rarely moves nav enough to be worth blocking the whole flush”
Official response
A community moderator stated that this is a known issue stemming from the most recent upgrade, that balances are growing even though this is not visible, and that a fix is being worked on. There has been no official response from Opentensor or the core team beyond this.
What to watch
Whether the fix for the price gate is deployed and whether the pending deposit backlog is resolved.

Voices on X

  • @opentensor

    Opentensor's weekly ecosystem recap highlights SN92 (Reax)'s REAX 1.3 model, which topped Hugging Face's Decision Index, beating over 270 competing models.

  • @ShizzyUnchained

    The account notes SN11's TrajectoryRL merged a default-on burn mode for miner rewards, distinguishing merged code from an executed on-chain burn, and flags changes at Zils, THOR, Vanta and Ormas.

  • @taodaily_io

    SN22 (Desearch) launched Open Complaint Radar, an open-source tool that tracks customer complaints on X and generates source-linked reports.

  • @TatsuEcosystem

    TatsuOS now accepts USDC and USDT alongside its native token for plans and AI credits on Testnet, ahead of a planned Mainnet promotion, with ChipForge's Quest 2 evaluator bundle finalized.

  • @TaoOutsider

    A community member reports an agent trading on SN91 (Ephemeris Cascade) forecasts grew from 5 τ to 5.1384 τ, a +2.77% gain, during a week described as rough for other subnets.

  • @CryptoZPunisher

    A community member asks SN74 (Gittensor)'s team about progress on roadmap milestones like GitHub bounty markets and code review agents, and whether it has generated external revenue.

  • @taodaily_io

    According to SubnetRadar data shared by taodaily_io, SN28 (SayGM) led 24-hour net buys at +760 τ, while SN44 (Score) led net sells at -580.02 τ.

  • @robotensor_ai

    SN58 (Attune Horizon) has launched, letting users teach robots using phone-captured data, after noting robots have learned from a small fraction of the text data used to train language models.

◆2026-11-09 · End of this edition

Recent editions

Past editions +