Bittensor Toolkit →Find the right tool for your question.

Bittensor Daily News

From the archive
#01 · SN51 · lium.ioverified
★★★★★

SN51 Lium reports $2.70 million in GPU spending over 31 days

Lium’s November 12 operating update reports $2.70 million in GPU spending for October 10 through November 9, compared with $1.45 million in the preceding 31-day period. Billed customers rose from 1,391 to 2,394, while rentals increased from 18,039 to 29,300. These are Lium’s reported operating figures, not independently audited financial results. The report puts 77.7% of usage outside the web application, up from 68.9%, and eight-GPU rentals at 5,536 versus 3,072. That points to a larger role for programmatic workloads and multi-GPU jobs within its reported activity. Its $34.6 million annualized spending run rate extrapolates the latest seven days; it is not realized annual revenue or profit. The post’s rental-growth percentage differs from its chart, so the absolute counts are used here.

Why 4 stars: Four stars: a substantial quantified change in actual reported usage matters more than a routine catalog update. Five is not established by one provider-reported period or an annualized extrapolation.

#02 · SN75 · Hippiusverified
★★★★★

SN75 Hippius highlights a public registry for models and containers

Hippius highlighted its Hub on November 12 as a distributed registry for AI models and containers. The public site was accessible during this review and listed 11,639 model entries and 1,766 containers, with formats including GGUF, safetensors, ONNX and PyTorch. These are registry counts, not unique customers or newly trained models. The team says developers can redirect huggingface_hub workflows by changing an import, while the registry exposes Docker, ORAS and Hippius tooling. The visible catalog includes Cascade model checkpoints and Reliquary-task containers. This verifies a usable public catalog and the team’s integration description, not an independent test of uploads, artifact delivery or decentralization guarantees.

Why 4 stars: Four stars: a substantive shared model-and-container distribution capability exceeds a minor feature. Five is not established because exceptional ecosystem-wide adoption is not shown.

#03 · SN71 · Leadpoetdeveloping
★★★★★

SN71 Leadpoet announces a Hermes plugin for outbound customer discovery

Leadpoet announced on November 11 that its customer-discovery workflow is available through a plugin for Nous Research’s Hermes agent. The team says users can supply a website, find companies showing buying signals, and have the agent draft and send outreach emails. Its follow-up provides installation and enable commands and a /customers entry point. This is Leadpoet’s product-availability claim, not confirmation of a commercial partnership with Nous Research. The installation path and end-to-end email delivery were not tested here, and no customer or revenue result is implied.

Why 4 stars: Four stars: putting lead discovery and outreach inside an agent workflow is a substantive product integration, beyond a routine model listing. Five is not established by this team announcement: a strategic partnership, deployment scale and customer outcomes are not demonstrated.

#04 · SN45 · AlphaRidge.aiverified
★★★★★

SN45 AlphaRidge publishes an event-discovery redesign

AlphaRidge’s November 11 version 4.0 documentation changes the subnet’s focus from analyzing supplied articles to finding events in finance, technology, politics and sports. Miners submit signed claims with source URLs; the coordination service records arrival, novelty and scoring, while validators check supporting pages and the ledger. The published rules distinguish a trusted-source discovery from an unconfirmed scoop. They describe a five-times reward for a scoop later confirmed by a trusted source and a two-times debt when it is not. This is a released mechanism and code change, not independent confirmation that every production validator has upgraded or that the approach improves news accuracy.

Why 4 stars: Four stars: a redesign of the subnet’s work and verification incentives is more consequential than maintenance. Five is not established by a code release without exceptional demonstrated reach.

#05 · SN117 · everyframe.studioverified
★★★★★

SN117 EveryFrame records three Gamma mints and three spends

Three successful Gamma mint-and-spend pairs for SN117 appear in the finalized interval reviewed today. A further mint and spend occurred on SN53 Engy. Seven subnets remain Gamma-enabled at the snapshot: SN25, SN53, SN58, SN80, SN81, SN103 and SN117. EveryFrame’s registered service describes converting Gamma into credits for image, video and audio generation. The transactions establish use of the payment mechanism, but wallet counts are not customer counts, and a spend event alone does not verify delivered output or external revenue. No subnet-to-subnet Gamma emission allocations were configured in the seven enabled entries at the cutoff.

Why 3 stars: Three stars: a concrete extension of observed usage to SN117 deserves coverage, but a handful of payments is not another network-wide launch or proof of commercial scale.

#06 · SN111 · Claimsverified
★★★★★

SN111 Claims raises the owner allocation and publishes miner-burn rules

A finalized SN111 allocation change at block 9,486,570 moves the owner share from approximately 18% to 26% and the miner share from approximately 41% to 33%, leaving validators near 41%. The five-slot vector assigns zero directly to the burn and recycle buckets. Separately, Claims published a validator policy directing 87.5758% of its miner-weight vector to an owner UID configured for burning and 12.4242% to miner hotkeys. Weekly winners are assigned 70%, 20% and 10% of that smaller miner pool. These validator weights are a different layer from the owner/validator/miner split. The document reports an encrypted commitment accepted at block 9,486,540; its eventual reveal and effective payout weights were not independently verified here.

Why 3 stars: Three stars: a meaningful subnet-specific distribution change, with finalized allocation evidence. It does not establish an ecosystem-wide protocol change or a completed transaction burn.

#07 · SN35 · Unknownverified
★★★★★

SN35 receives a new registration after the previous subnet is removed

The chain removed the previous SN35 registration at block 9,485,498 and added a replacement at block 9,485,521. The new registration has different owner keys and no published subnet identity at the reviewed cutoff. The queued registration recorded 85.943564944 TAO locked; that is a protocol registration amount, not evidence of a negotiated purchase price. The new operator’s project name and public channels remain unresolved. Historical price comparisons across the replacement are invalid, and the website now excludes the crossing 24-hour and seven-day returns. A new registration does not establish the identity of its beneficial owner or continuity with the former project.

Why 4 stars: Four stars: replacement of the registered subnet and its controlling keys materially changes what occupies SN35. Five is not supported; the new project and operating scope remain unknown.

#08 · SN8 · Vantaverified
★★★★★

SN8 Vanta records five more qualifying alpha burns

Five successful SN8 AlphaBurned events destroyed 133.333333333, 225.004624293, 172.276442599, 207.546885395 and 139.895298531 alpha in the reviewed interval. Each exceeds the archive’s 100-alpha per-event threshold. These are transaction burns, kept separate from emission withholding, recycling and fees.

Why 3 stars: Three stars: verified token destruction is a useful economic update, but these individual burns do not establish a major change in the subnet’s operating model.

#09 · SN36 · Weirverified
★★★★★

SN36 Weir publishes its GPU miner agent and connection procedure

Weir published a miner agent and revised setup documentation on November 12. Operators register control endpoints privately with Weir support, and validators check GPU authenticity and performance. The guide describes a 14-day record of passed checks for idle-reward eligibility. This clarifies how outside machines enter the GPU rental system. It is not permissionless self-onboarding: coordination with the operator remains part of the documented process, and live onboarding was not tested in this review.

Why 3 stars: Three stars: concrete participation tooling with a bounded onboarding consequence. Four is not established as a new marketplace launch.

#10 · SN116 · Carbonverified
★★★★★

SN116 Carbon explains decaying rewards for winning physics models

Carbon published its miner-emission rules on November 12. A qualifying leading model receives its challenge’s full share for 24 hours; without a further qualifying improvement, the rate halves each day and the unpaid portion is burned. A verified improvement can restart the reward period, while a challenger can replace the incumbent. The article’s constant-emission example approaches two full-share days of cumulative payment from a single win. These are competition rules, not an income forecast or proof of mainnet payments. The article describes four equal launch shares, while the reviewed repository describes two initial challenges; the exact launch allocation remains unresolved and is not presented as settled.

Why 3 stars: Three stars under the competition cap: useful incentive rules, with conflicting launch-share descriptions explicitly unresolved.

#11 · Ecosystemverified
★★★★★

Taostats publishes a practical guide to multisig and proxy wallets

Taostats’ November 11 guide explains how its Actions interface creates and uses multisignature wallets and restricted proxies. Multisig requires a chosen signing threshold, while proxy types limit which operations a delegated wallet can perform. The article links the relevant setup and transaction tools. The practical distinction is between separating approvals and delegating limited day-to-day work. Neither removes risk: losing enough signers can lock a multisig, and a staking proxy can still incur trading losses. This is an educational guide to existing tools, not a newly deployed chain security feature.

Why 3 stars: Three stars: timely, concrete operational guidance for wallet users. Four would overstate a guide to existing functionality as a new security capability.

#12 · SN94 · Cathedralverified
★★★★★

SN94 Cathedral adds code for persistent encrypted workspaces

Cathedral published a sealed-data-volume implementation that lets workspace data survive a restart of the same guest build. It uses a separate encrypted volume instead of placing persistent work on a scratch disk whose key changes at each boot. The release explicitly says this phase is not deployed on a live host. Persistence across a new build remains outside this phase. The practical advance is implementation of restart persistence under the documented threat model, not an independently audited security guarantee.

Why 3 stars: Three stars: a substantive implementation milestone with clear limits. Four as a live capability is not supported because the release explicitly says deployment has not happened.

#13 · SN89 · InfiniteQuantverified
★★★★★

SN89 InfiniteQuant adds runner signatures to agent-book receipts

InfiniteQuant published a requirement for an allowed book runner to countersign receipts received from November 11 at 22:00 UTC. The signature binds the receipt to an algorithm hash, data watermark and book-state hash. Earlier receipts are left unchanged. This is a code-level validation rule; deployment across all validators was not independently checked.

Why 3 stars: Three stars: a meaningful change to which trading-agent receipts qualify. Its bounded validation scope does not establish a major product launch.

Smaller developments and early plans. Verification labels distinguish confirmed information from developing claims.

#14 · SN17 · 404—GENverified
★★★★★

SN17 404 improves recovery when validation workers stall

404 published changes that retain already-collected validation outcomes when a later checkpoint cannot be read and replace workers that stop progressing for 45 minutes. The code also allows preferred data centers to be configured. These are reliability improvements in the evaluation pipeline, not a change to the competition winner or evidence of faster production generation.

Why 2 stars: Two stars: a bounded reliability repair rather than a new product or scoring mechanism.

Stars indicate editorial priority, not investment quality. Routine competition updates receive at most three stars. Shizzy videos receive three; substantive non-video Shizzy coverage receives at least four. Scheduled events remain plans.

Sources & coverage notes
04

Audit trail

Coverage notes

SN51 Lium reports higher GPU spending and customer activity. SN75 Hippius shows its public model registry, SN45 AlphaRidge changes event discovery incentives, and SN117 EveryFrame records Gamma payments.

Coverage: Nov 11, 2026, 12:40 PM – Nov 12, 2026, 9:40 AM ET

Searched

  • All 128 current subnet identity/website records; authenticated public repository review; bounded native team and ecosystem timelines; owner bookmarks and recent posts/replies; original Lium media and selected primary product/policy sources.
  • Contiguous finalized lifecycle, emission-allocation, qualifying burn and Gamma event review; owner holdings refreshed at one recorded finalized block.
  • Actual signed-in Grok discovery and qualitative attention queries; final omission query completed; uncorroborated suggestions retained as unresolved leads.

Public-data gaps

  • Discord signed out during this run; official team-channel and price-move investigations remain blocked until access is restored.
  • X profiles and public repositories were reviewed in bounded windows. Complete seven-day Posts/Replies/Reposts and all attached media were not reviewed for every subnet or fund principal; unresolved leads remain pending.
  • Some official websites, repositories and most secondary briefs were inaccessible. These access gaps are not findings of no news.
  • SN35’s new project identity, Carbon’s conflicting launch-share descriptions, and the cause of several large token moves remain unresolved.
  • The full site audit is still in progress. Tested market repairs do not establish that every route and interaction passed.
Browse all 100 editions

Conference news · September 30, 2026

Exploit Conference Catch Up · Day 1

The teams, the ideas and the practical takeaways from Exploit Summit’s first day in Montreal. From Gamma and phone-based AI to robotics, science and the business of running a subnet.

Catch up on Day 1

Conference news · October 1, 2026

Exploit Conference Catch Up · Day 2

Catch up on Exploit Summit’s second day: Carbon’s Pitchtensor win, SN107 Minos’ research app, agent controls, robotics data, inference partnerships and the decisions behind building a subnet business.

Catch up on Day 2