TAO (TAO)

A Bittensor subnet just shipped a fake 120B model and got caught in under 48 hours. Alpha fell as much as 75% intraday. Most people read that as the death of Subnet 24. It is the strongest proof of what $TAO is actually building. Quasar sold the dream of crushing the long context barrier. Linear scaling memory. A 120B model to prove it. Then independent analysis pulled the checkpoint and found something ugly. 96 to 98% weight match with Ant Group's open source Ling mini base 2.0. The novel architecture claims did not exist in the file. Loop Transformer, Engram, all of it missing from the actual weights. The original credits were stripped. That is the whole story in one line. The team thought open weights were a distribution channel. Open weights are a lie detector. This is why verification beats reputation on a network like Bittensor. A centralised lab buries a reskin behind an API and a launch thread forever. Here anyone could pull the receipts, and the market repriced trust in hours instead of yea

Bittensor today. 24 of the 121 subnet team channels we follow had news, plus the ecosystem press. SN24 Quasar: Model Plagiarism Allegations Crater Subnet 24 SN5 Hone: Upcoming Dashboard and Problem Difficulty Increase The Hone team confirmed a dashboard is nearing completion with a timeline update promised for Monday. SN118 Ditto: Implements Scheduler Fixes and Screening Process Updates The team resolved a scheduler issue blocking position 6 assignments and introduced automated screening improvements. SN11 TrajectoryRL: Discusses Score Decay and Site Outage TrajectoryRL (SN11) experienced a brief live site outage resolved by Crabby_tt. SN102 ConnitoAI: Validators Failing to Assign Scores to Miners Validators on ConnitoAI (SN102) are not assigning scores to miners for multiple cycles. From the press: What Are Bittensor Subnets? (https://t.co/4NtwMGnMF8) This Week in Bittensor: The Emission Gate Starts Cutting (TAO Protocol) Exploit Summit 2026: How Bittensor’s Biggest Community Event Came Together

TAO dips keep getting bought I don’t know when it breaks, but every dip is getting absorbed. Bulls are clearly taking control. Nothing is guaranteed, but once this moves, late buyers may not get much time. Feels like a matter of when, not if. bittensor:native https://t.co/IYuD3zdRW3

@markjeffrey I'm bittensor:native pilled now. Just bought some today.

imagine backing a bittensor subnet for months, only for an analysis to allege that its flagship model was 96% identical to one already released elsewhere. SN24 aka quasar crashed over 70% as the claims spread. tldr: → quasar is subnet 24 on bittensor → it claimed its 18B model was trained on more than 1 trillion tokens → the project raised 400 $TAO and was publicly backed by @const_reborn → an analysis later compared the model with one released by InclusionAI → the result reportedly found a 96% match → using an open-source model was never the problem → the issue is whether quasar presented an existing model as original work → once the report spread, holders rushed to unstake and sell → SN24’s shallow liquidity made every exit push the price lower → and boom, quasar crashed over 70% within hours → now people are questioning the team, its model claims and everyone who vouched for it I honestly don’t know what’s worse here: that claims this big went unchecked for months, or that so many of us treat

There are maybe 10 Series A caliber startups in Bittensor and 3‑5 at Series B. Everything else is a pre‑revenue seed stage moonshot. Size accordingly as your base case is zero, but your upside is 100x or more. Founder is everything, and you need a long time horizon to let them cook. TAO is precious. Spend it wisely.