BITTENSOR · THE NEWS ARCHIVE
Search 88 dated editions without scrolling through them one by one.
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 A past verified announcement can now be outdated.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.
Affine's August 13 brief records discussion about protecting model privacy while limiting spam. No new public repository commit landed inside the cutoff, so this remains a team-reported design discussion rather than a shipped successor to yesterday's benchmark-publishing work.
Subtensor main merged a runtime change that scales each subnet's EMA-price share by one minus its MinerBurned proportion, renormalizes, and then applies the emission gate. Withheld miner incentive counts whether recycled or burned. The commit bumps the development spec to v445, but a merge to main is not evidence that Finney has enacted the runtime.
At the delayed snapshot, Chutes led displayed emission at 12.09%, followed by lium at 10.37%, Minos at 8.27%, Affine at 8.25% and Targon at 7.48%. Ditto showed the largest one-hour increase, rising from zero to about 0.82%, while gm fell roughly 0.19 percentage points and Targon about 0.05 points.
Between the August 12 report roster and the latest indexed snapshot, SN39 changed from Cathedral to deprecated and SN108 changed from TalkHead to Prometheon; SN54 shortened from Yanez MIID to Yanez. The public metadata change is verified, but it does not by itself prove a registration, deregistration, ownership transfer, or project acquisition.
At the delayed 12:51 PM ET snapshot, 46 of 128 directional subnet markets were positive and 82 were negative, with about 138,751 TAO of displayed 24-hour volume. Leadpoet gained 49.1% while DSperse fell 18.6%. SubnetRadar separately detected six historically profitable wallets buying about 254 TAO of Leadpoet over four hours; that wallet classification is provider-generated, not proof of coordinated intent.
OpenRoboto added about 339 TAO to its pool over the latest hour and Vidaio added about 233 TAO. RedTeam lost about 314 TAO, OxMarkets about 114 TAO, Mainframe about 93 TAO and Synth about 92 TAO. These are reserve changes observed after the scheduled cutoff because publication was delayed; they do not establish wallet motive.
Hone announced a breaking v3 release adding explicit Python and Rust problem support and setting owner burn to zero. Coordinated miner and validator upgrades are required; this was an announced deployment at the report cutoff, not yet a completed rollout.
Iota said a drained-wallet incident is under investigation and that infrastructure security is being rotated. Orion 16B payouts moved to manual processing while the incident response continues.
Trishool introduced one reusable jailbreak template evaluated across multiple objectives, alongside a guard-model update and scoring reset. The format is live according to the team brief; performance outcomes remain to be measured.
Score will randomly publish daily audit samples from top private-track performers, including miner responses and evaluation evidence. The public repository also merged per-element audit storage work during the window.
Nepher explained that small conviction-lock adjustments could appear to have no effect because perpetual and decaying lock buckets interact during roll-forward. The team described the issue as a display/accounting-path bug, not a token cliff.
Following the previous day's backlash, Engy described two paths: permissionless TEE onboarding for supported frontier models and a vetted enterprise track. The team also announced a buyback program. The policy direction is confirmed; execution and buyback scale remain team-reported.
NIOME said v3 is moving from testnet to mainnet this week after workflow stabilization. The team said emissions will reward top-performing miners; the exact activation block was not published at the cutoff.
Bitsec identified a fingerprinting failure affecting validator scoring and said a provenance fix would ship before the next round to make simple copycat strategies less effective.
Chutes restored Kimi K3 after a miner outage. Users continued reporting material performance variance across several models, so availability was restored but quality concerns remained open.
BASE merged a fix for weights being sealed against results from a different epoch and instructed miners to update. This is a required participant update tied to scoring correctness.
Beam discussed expanding rewarded miners from the top 20 to the top 30 and clarified that PRISM bandwidth scoring is constrained by the slowest chunk in a batch. The change broadens participation while retaining performance gates.
A developer challenged hotkey restrictions visible in Minotaur's code as unfair; the team defended them as anti-spam enforcement. The policy is visible and disputed, but the surrounding allegations were not independently adjudicated.
Ditto activated benchmark v9 but found bugs requiring another test cycle. The team is rerunning leading v8 agents to trigger v9 incentives while validator instability continues to delay scores.
Affine's public repository added per-task benchmark rollout records, trajectory publishing to Hippius, and a metrics page during the window, providing verifiable development activity beyond team chat.
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.