Dates and verification labels belong to the original edition. A past verified announcement can now be outdated. Historical project names are retained; an SN filter searches the slot across registrations. All keywords must match the story text or subnet label. Ratings describe editorial priority, not investment quality.
SN51 Lium reports 225 B300 GPUs and 91% rented supply
SN51 Lium’s October 22 update says its marketplace hosts hardware valued above $30 million, including 225 B300 GPUs, with 91% of supply rented. These are team-reported inventory, valuation and utilization figures, not independently audited revenue or profit. They provide an operating measure beyond token trading. Hardware valuation is not subnet market capitalization, and the percentage is a point-in-time statement rather than a guarantee of sustained demand.
Specific inventory and rented-supply figures offer useful adoption context beyond price, while remaining team-reported.
SN55 NIOME acknowledges storage failures affecting uploads
SN55 NIOME’s October 22 digest attributes upload failures and skipped miners to validator storage limits, with a fix in progress. The repository has retry-logic work, but that does not prove the storage issue resolved. The digest also says there is no miner cap or whitelist. This remains an attributed availability issue rather than evidence of deliberate exclusion.
Upload failures and skipped miners are consequential operating problems with no independently established resolution.
SN58 Attune closes Vector Level 1 and opens Level 2
SN58 Attune’s official October 21 results report 87.9% success across sixteen dual-arm tasks in unseen scenes from one demonstration, versus 87.2% for its fine-tuned pi0.5 baseline. The short media shows competition progress; an intermediate frame differs from the final text, so figures here are attributed to the written announcement. The team says Level 2 is live, repeated by co-founder R.L. Bryer later that day. These are benchmark and participation results, not independently verified commercial deployment.
Competition results and the next level are useful participation developments; they do not establish commercial robot adoption.
SN62 Ridges reports two unstable validators disabled
SN62 Ridges’ October 22 channel digest reports two validators disabled because of instability, with investigation ongoing. It does not establish completed recovery, lost funds or a precise outage interval. The availability issue is retained with attribution while a primary public incident report remains unavailable.
Disabling validators is a material availability issue for participants; recovery remains unverified.
SN66 Conjectures publicizes a restricted four-speed lonely-runner proof
SN66 Conjectures’ October 22 announcement links a Lean-accepted proof for a restricted amended lonely-runner spectrum statement: four distinct positive integer speeds, overall gcd one and pairwise gcd at most two. Its verification record reports an unchanged target statement, permitted axioms and an accepted Lean replay; a second independent kernel was not run. We inspected the record and linked solution, but did not rerun the proof or establish novelty. This concerns the stated restricted case, not the full lonely-runner conjecture.
A public restricted theorem and validation record provide inspectable scientific output; novelty and independent replay remain unverified.
SN66 Conjectures adds a float-kernel optimization challenge
SN66 Conjectures’ October 22 challenge asks miners to optimize floating-point kernels while proving outputs match FloatLib and TorchLean specifications. The team reports its champion at 8.7 times the shipped reference on its benchmark. That is a competition-specific claim, not a general hardware benchmark or independently rerun result. The participation opportunity combines speed optimization with a formal correctness condition.
The challenge rewards formally constrained kernel optimization; the speed claim remains benchmark-specific.
SN67 Harnyx removes its paid-model allowlist
SN67 Harnyx’s October 21 account update says miners can use any paid OpenRouter model instead of a team-selected allowlist. Regular pricing and task budgets remain; free or promotional pricing is not a route around those limits. This expands choices in the existing research competition, rather than establishing a new performance result.
Removing a paid-model restriction expands miner choices within the existing competition, without a new performance claim.
SN80 OpenRoboto plans Solana availability through ForeverMoney
In an October 22 announcement at 11:56 a.m. ET, bridge provider ForeverMoney said SN80 OpenRoboto will become available on Solana through its bridge, with Chainlink CCIP powering the planned cross-chain connection. It also plans Solana bridging for TAO and other subnet tokens. The post uses “will go live” and “once live”: it is an announced rollout, not confirmation that Solana transfers are already available. ForeverMoney describes SN80 as the first subnet to span Base, Robinhood Chain and Solana; that first-of-kind claim and production availability have not been independently established. No launch time or token contract was provided in this post.
A named bridge provider announces a concrete additional-chain distribution plan for SN80; completion and first-of-kind claims remain unverified.
SN85 Vidaio schedules Competition 004
SN85 Vidaio’s October 22 digest announces enrollment October 23–31 and evaluation from November 1 for video-upscaling Competition 004. The task restores downscaled video with quality metrics and a speed component. These are upcoming participation dates attributed to the digest; no winner or completed quality gain is being announced.
Dated enrollment and evaluation windows are actionable competition information, not a completed product result.
SN92 REAX details bug remediation and Friday rules
SN92 REAX’s October 22 digest reports a capacity-probe bug affecting roughly forty hotkeys during October 21, 15:30–17:04 UTC, with compensation planned. Public release code removes late-check stale-answer rejection and prepares larger generalist draws plus per-coldkey diminishing returns in small pools. Next rules are scheduled for October 23; merged code does not prove universal activation or completed compensation. Const’s favorable October 21 comment concerns the already-announced REAX 1 comparison with its Perplexity base, not an independent new benchmark.
Bug remediation and scheduled rule changes materially affect miner evaluation and rewards; compensation is not confirmed complete.
SN97 Ingot’s FP8 result draws Const’s endorsement
Const’s October 21 post says SN97 Ingot’s Qwen 27B FP8 quantization improved on Alibaba’s version after a week of competition. This is an attributed comparison, not an independently reproduced benchmark. The current project site marks its eight-bit round closed and two-bit round open, so that next target is already reflected in its interface. Const’s follow-up that all models will be trained in markets is commentary. The inspectable compression workflow matters more than an unsupported claim of superiority on every task.
A reported quantization comparison and open two-bit round are useful competition developments, not independently reproduced superiority.
SN107 Minos’s first ASHG poster session is scheduled today
Reminder from the previously reported team schedule: SN107 Minos’s synthetic whole-genome-cohort presentation is scheduled for October 22, 4:15–6:15 p.m. Eastern in Montreal, Exhibition Hall Room 220. Its variant-calling benchmarking session follows October 23, 2:30–4:30 p.m. Eastern. At this report’s cutoff both remain upcoming. The official event site confirms the meeting; these precise slots remain team-reported. No completed presentation, new customer or commercial OpenAI relationship is inferred. The team’s separate IM2 release date is still pending.
A time-specific reminder is useful for attendees, but repeats a scheduled event rather than establishing scientific adoption.
SN114 SOMA reports higher token savings with Jev
SN114 SOMA’s October 22 update reports token savings increasing from 30% to 40% after its first competition using Jev. Earlier integration coverage connects its context-compression workflow to SN28 SayGM, SN53 Engy and SN64 Chutes. These are team-reported savings dependent on workload and evaluation choices, not a guaranteed reduction for every user. Developers should check preserved output quality alongside reduced tokens.
Workload-dependent token savings are useful team-reported competition output, with quality preservation still to assess.
SN117 Everyframe publishes epoch-accounting recovery tools
SN117 Everyframe’s validator v0.5.0 adds canonical epoch accounting and tools for assignment publication, epoch finalization and reward recovery, following yesterday’s scheduling work. The channel digest includes member reports of an RPC-related payout backlog, but no confirmed team resolution or independently reconciled total. Published recovery tools are concrete; settlement of every delayed reward remains unverified.
New epoch-accounting and recovery commands address reward reliability, without proving every delayed payout settled.
SN117 Everyframe raises owner allocation to 33%; validator/miner shares fall
Late chain update: at 22:44:48 UTC on October 22 (6:44:48 p.m. ET), SN117 everyframe.studio’s owner successfully changed its eTAO alpha-emission split at finalized block 9,342,125, extrinsic 16. Owner / validators / miners changed from 18% / 41% / 41% to 33% / 34% / 33%. The exact five-slot vectors were checked against parent-block and execution-block storage; burn and recycle slots remain zero. The validator bucket includes Root, which receives its live baseline first subject to the validator-allocation cap. Historical staker APR/APY may lag as rewards from the new split enter the reporting window; this does not establish a specific yield or price outcome. This verified supplement was reviewed after the original daily research cutoff.
A finalized owner action materially redistributes rewards: owner +15 percentage points, validators -7 and miners -8. Exact storage and successful execution verified.
SN118 Ditto updates custody compatibility for runtime 482
SN118 Ditto’s v0.381.2 source adds a Finney 482 runtime interface and recovery compatibility for its treasury workflow. This application-level response is distinct from its proposed bounty system, whose payout and hotkey connections remain unfinished in channel coverage. Merged code does not establish that particular funds moved or that every deployment is updated.
Runtime-482 compatibility is concrete operating work; it does not establish fund movement or finished bounty payouts.
Finney runs runtime 482 with coldkey-swap and proxy changes
The v482 release was published October 21 at 19:59 UTC, and direct finalized metadata confirms runtime 482 is active. Release notes cover merging coldkey swaps into used destinations, Validate and Weights proxies, proxy fee payment through three levels, Yuma payout handling and a Null-consensus scheduling fix. TaoStats says Actions has switched to announceColdkeySwapV2. These are protocol and compatibility changes, not a fresh owner emission-allocation transaction for every subnet. Wallet and operator integrations should check call compatibility.
A chain-confirmed runtime upgrade with material wallet, validator and proxy compatibility consequences.
Shizzy’s October 21 members news video is available
The latest members update is the October 21 Bittensor news and price-action episode, approximately eight minutes long. YouTube marks it members-only, so playback remains through membership access. The existing public default is preserved; the already-covered SN58 Attune interview remains a replay rather than a newly scheduled stream.
Owner editorial rule: Shizzy video coverage is three stars; membership access remains clearly labelled.
SN71 Leadpoet starts a sales pilot with SN10 Pareton
SN71 Leadpoet and SN10 Pareton both announced their pilot on October 22. The work is to identify AI companies facing GPU-cost and inference bottlenecks, find technical decision-makers and run targeted outreach for SN10 Pareton. This is a concrete trial of the sales-agent service beyond its earlier launch. Neither team disclosed a contract value, customers won or resulting revenue; the pilot should be judged on subsequent conversion results.
Matching primary announcements establish a concrete customer-development pilot, with revenue and conversions undisclosed.
SN1 Apex submission-timeout reports remain unresolved in available sources
SN1 Apex miners reported submission timeouts and 504 errors for a competition round in the October 21 channel digest, including paid attempts that did not appear on the board. The digest does not establish a team resolution or a complete outage interval. Separately, the public NIST assembly competition code publishes v0.3.0 observation hardening and pinned images. That code release is not evidence that the reported submission issue is fixed. Miners should distinguish competition-rule updates from the availability of the submission service.
Relevant operational or planned change, but primary execution or full incident details remain unverified.
Searches published headlines and summaries, not private bookmarks, unpublished leads or source-page contents. Duplicate headlines within an edition appear once; separate editions remain separate so follow-ups are preserved. Share the page URL to retain your filters.
