Bittensor Toolkit →Find the right tool for your question.

BITTENSOR · THE NEWS ARCHIVE

Find the story.

Search 86 dated editions without scrolling through them one by one.

Reset filters

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 A past verified announcement can now be outdated.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.

2,335 matching stories

2026-10-28 edition →SN16 · end gameverified2 / 5 stars

SN16 changes its name from Zero Hour to end game

Successful SubnetIdentitySet at finalized block 9380213 changes Zero Hour to end game. The description, GitHub, website and contact fields are empty at the reviewed cutoff. A robotics pivot, new product or sale cannot be inferred from an aggregator label. Historical Zero Hour stories retain their original names.

Executed identity update without unsupported business conclusions.

2026-10-28 edition →SN17 · 404—GENdeveloping3 / 5 stars

SN17 404-GEN reports client and SDK changes

The dated brief reports a witness-client fix for a /v1/ request error, an async SDK, old-SDK deprecation and reshuffled submissions against a base benchmark. Original private messages and the exact upgrade deadline were not independently accessed; check pinned operator instructions.

Concrete participation, scoring or service-status information; secondary details remain developing.

2026-10-28 edition →SN20 · Witnessverified3 / 5 stars

SN20 Witness: finalized allocation carried into today’s report

The full-window review verifies the successful allocation event at block 9378110. Owner/validator/miner shares are approximately 33%/7%/60% of the subnet allocation, versus the previous 18%/41%/41%. Exact before/after five-slot vectors and historical registration labels remain in the eTAO archive. Allocation settings are not transaction burns, guaranteed APY or a specific validator’s realized return; yield series can lag the change.

Verified economic settings; carry-forward does not count a second event.

2026-10-28 edition →SN22 · Desearchverified3 / 5 stars

SN22 Desearch publishes Topic-to-RSS example

The October 28 example turns a query into a feed of dated results, filters undated pages, deduplicates URLs and titles, and sorts newest first. It is a pattern for agents following topics without a native feed. The announcement and example link were reviewed; long-running reliability was not independently tested.

A concrete reusable developer search workflow.

2026-10-28 edition →SN24 · ms.computerverified2 / 5 stars

SN24 changes its name from Quasar to ms.computer

Successful SubnetIdentitySet at finalized block 9378507 changes Quasar to ms.computer and updates the website field. Full-window owner and registration comparisons found no executed ownership/registration change for this SN. Discord channel-claim commentary does not establish a subnet sale. The large dated alpha/TAO move does not independently establish its cause or a product launch.

Executed identity change, separated from ownership and price causality.

2026-10-28 edition →SN28 · SayGMverified4 / 5 stars

SN28 SayGM adds SN64 Chutes as a confidential inference provider

The October 28 integration gives SN28 SayGM users another upstream for confidential open-weight inference. It does not mean every proprietary-model request is executed by SN64 Chutes, or that every model has the same privacy tier. Documentation distinguishes confidential -tee models and provider selection; check the current catalog and privacy policy. No customer volume or audited savings were established in this review.

A documented new provider connection that changes gateway access.

2026-10-28 edition →SN33 · ReadyAIverified3 / 5 stars

SN33 ReadyAI: finalized allocation carried into today’s report

The full-window review verifies the successful allocation event at block 9376469. Owner/validator/miner shares are approximately 18%/49%/33% of the subnet allocation, versus the previous 18%/41%/41%. Exact before/after five-slot vectors and historical registration labels remain in the eTAO archive. Allocation settings are not transaction burns, guaranteed APY or a specific validator’s realized return; yield series can lag the change.

Verified economic settings; carry-forward does not count a second event.

2026-10-28 edition →SN34 · Yugosdeveloping3 / 5 stars

SN34 Yugos resets history and pauses submissions

The dated brief reports a v0.3.0 champion-history reset so routing-only entries do not inherit old multi-call results, and a pause for incentive/private-dataset changes. Published docs explain wording-gate enforcement. This is not an independent fairness audit.

Concrete participation, scoring or service-status information; secondary details remain developing.

2026-10-28 edition →SN38 · ChronoLLMdeveloping2 / 5 stars

SN38 owner announces a coldkey swap; execution is pending

Finalized block 9381551 records ColdkeySwapAnnounced for the current SN38 owner. The event field is new_coldkey_hash: 0xf32c…faa5 is a commitment, not a verified destination address. Owner storage remains unchanged; the pending tuple points to block 9417551. This is a delayed key-migration announcement, not a completed subnet transfer, identified buyer or sale price.

Verified pending key operation with hash/address confusion resolved.

2026-10-28 edition →SN42 · Unknownverified2 / 5 stars

SN42 sets an owner operating hotkey

Finalized block 9376369 records a successful SubnetOwnerHotkeySet for SN42. This updates an operating role, not the registered owner coldkey. The full-window owner/registration comparison found no SN42 ownership or registration change; no sale or named new team is inferred.

Verified operational event without a claimed ownership transfer.

2026-10-28 edition →SN44 · Scoredeveloping3 / 5 stars

SN44 Score reports annotation-error removals

The dated brief reports five ground-truth-error blocks removed across PPE, Petrol Station and Fallen Person, with edge cases retained. This changes scoring inputs, not proof of a model-accuracy breakthrough. Private annotation records were not independently audited.

Concrete participation, scoring or service-status information; secondary details remain developing.

2026-10-28 edition →SN44 · Scoreverified3 / 5 stars

SN44 Score demonstrates people and vehicle counting at Shibuya

The October 28 video overlays pedestrian and vehicle detections and a running count on Shibuya Crossing footage. It illustrates dense-scene computer vision beyond a text announcement. The displayed counters are demo outputs, not independently scored accuracy results; the clip does not establish a new paid deployment or a public production launch.

A concrete visual demonstration; commercial deployment and accuracy are not inferred.

2026-10-28 edition →SN46 · Instantdeveloping3 / 5 stars

SN46 Instant reports dashboard work and an overloaded API

The dated brief says a swarm test overloaded the portal/API while revenue/usage data and token-counter visibility were being corrected. Recovery was expected by morning, not independently verified here. Launch and recovery remain developing; planned charts do not establish audited revenue.

Concrete participation, scoring or service-status information; secondary details remain developing.

2026-10-28 edition →SN50 · Synthdeveloping3 / 5 stars

SN50 Synth links miner rewards to AutoQuant benchmark performance

SN50 Synth’s October 28 announcement describes a collective performance benchmark built from frontier-agent forecasting models: greater miner outperformance reduces the miner-emission burn and leaves more rewards for miners. The team reports 22% for this week, unchanged from last week, and links its open-source AutoQuant work. This affects the stated incentive mechanism; it is miner-emission withholding, not a qualifying transaction burn, an alpha-sale event or guaranteed miner earnings. The live reward calculation and benchmark outcomes were not independently reproduced.

A concrete miner-incentive update, with the unchanged weekly withholding rate distinguished from transaction burns.

2026-10-28 edition →SN51 · lium.ioverified3 / 5 stars

SN51 lium advertises B300 spot capacity at $4 per GPU-hour

SN51 lium’s October 28 post advertises B300 GPUs at $4 per GPU-hour and reports 550 B300s listed, with 443 rented. The attached listing is marked SPOT and shows $32 per hour for the displayed machine ($4 per GPU), so the headline is not the price of the whole machine or a guarantee of uninterrupted reserved capacity. Inventory, rentals and pricing are the team’s dated figures, not independently audited revenue or a guaranteed current quote.

Actionable compute pricing with per-GPU, machine and spot terms distinguished.

2026-10-28 edition →SN54 · Yanezverified3 / 5 stars

SN54 Yanez: finalized allocation carried into today’s report

The full-window review verifies the successful allocation event at block 9376080. Owner/validator/miner shares are approximately 18%/49%/33% of the subnet allocation, versus the previous 18%/41%/41%. Exact before/after five-slot vectors and historical registration labels remain in the eTAO archive. Allocation settings are not transaction burns, guaranteed APY or a specific validator’s realized return; yield series can lag the change.

Verified economic settings; carry-forward does not count a second event.

2026-10-28 edition →SN62 · Ridgesdeveloping3 / 5 stars

SN62 Ridges starts a baseline phase and signals Ridgeline access

The October 27 team post describes a benchmark phase where miners compete to set the initial leader without emissions; normal rewards follow. The October 28 brief says parts of Ridgeline open within the week. Access remains planned, not a verified general launch.

Concrete participation, scoring or service-status information; secondary details remain developing.

2026-10-28 edition →SN64 · Chutesverified4 / 5 stars

SN64 Chutes confidential models arrive on SN28 SayGM

Both teams announced the integration on October 28. Builders can access SN64 Chutes-served open-weight models through the SN28 SayGM gateway alongside its other providers, using one key and base URL. The technical article identifies confidential models by a -tee suffix and documents X-GM-Providers: chutes to pin routing. This expands distribution across two subnet products. The teams describe TEE protection at gateway and inference layers; this review did not execute a paid request or independently audit attestation and privacy guarantees.

A concrete cross-subnet product integration confirmed by both teams and technical routing documentation.

2026-10-28 edition →SN65 · True Performance Networkverified3 / 5 stars

SN65 True Performance Network announces base-versus-quant comparison

The October 28 Bench announcement describes next-token distribution divergence, hidden-layer comparison and cross-model perplexity. These help developers examine whether a compressed model behaves like its source beyond a headline score. The team links the Workbench; the attached screenshot shows Quantize, so the image alone does not verify Compare results. No independent model pair was benchmarked here.

A useful developer capability with bounded launch and testing evidence.

2026-10-28 edition →SN68 · NOVAdeveloping3 / 5 stars

SN68 NOVA develops a Hit-to-Lead agent with SN64 Chutes and OnePot AI

In its October 28 announcement, METANOVA describes extending a NOVA Blueprint winner’s algorithm from finding molecular hits into iterative lead optimization. OnePot AI supplies proprietary chemical databases; SN64 Chutes separately confirms collaboration and support for the developing agent. METANOVA says domain-specific models were integrated for confidential TEE inference, with integration into its own development program still the next step. This is a development collaboration, not a completed drug, clinical result, or independently validated synthesis or privacy outcome.

A specific drug-discovery workflow and confirmed infrastructure collaboration, with program integration still pending.

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.