xxxxxx

[Pushingτao (τ,τ) @PushingTao]

Compute costs will come down significantly as AI scales. When you combine that with compute subnets like @chutes_ai and the Bittensor incentive mechanism a new world of possibilities opens.
We're already getting to the stage where datacenter level hardware can sit on your desk. AMD's recent announcement is a perfect example of this.
Now imagine individuals worldwide running hardware like this, contributing compute to Bittensor subnets and earning emissions back in return.

A distributed network of local datacenters. We're closer to that world than most people think.
$TAO $dTAO

[Openτensor Foundaτion @opentensor]

Bittensor Ecosystem Highlights :: June 8–14, 2026

SUBNET ACHIEVEMENTS

[ @chutes_ai - SN64 ]
@jon_durbin shared a draft of the Parallax tech report, outlining a MoE training method to reduce per-participant VRAM and FLOPs.
> https://t.co/KJzo9YGfcB

Chutes also became a launch partner for Respan’s new AI Gateway.
> https://t.co/pBhdJWkKYj

[ @QuasarModels - SN24 ]
Quasar released Quasar-Preview, its first public Quasar model trained on Bittensor: 18B MoE, 2B active and 5M context.
> https://t.co/ThEA0gpgzg

Quasar is preparing a 10T-token decentralized training run on SN24, starting with a 5T-token phase to produce a stronger checkpoint.
> https://t.co/vXiQdA7fHK

[ @oroagents - SN15 ]
ORO shared its arXiv pre-print, code, data and post-training pipeline for building shopping agents from SN15’s open agentic shopping traces.
> https://t.co/rICbXX6vj4

[ @webuildscore - SN44 ]
Score showed how its 19MB vision model beat larger AI models on object detection while running much faster on CPU.
> https://t.co/kfAJ78OmR8

They also added new comparison pages on @manakoai against ChatGPT, Claude, Roboflow, SAM 3 and other vision AI tools.
> https://t.co/D5ARXilc8j

[ @vidaio_ - SN85 ]
Score is partnering with Vidaio to bring vision AI challenges to SN44 and make video archives searchable and actionable.
> https://t.co/bRHvHIhw8R

[ @yanez__ai - SN54 ]
Yanez partnered with Nexartis, an identity and trust infrastructure company, to help verify human, AI model and agent activity across digital transactions.
> https://t.co/RjqUsoyDUW

They also shared in their latest AMA that Yanez has generated $300K+ in 2026 sales, with an active pipeline over $1M and 11 clients.
> https://t.co/yHaFxPX5Oy

[ @trishoolai - SN23 ]
Trishool was accepted into Anthropic’s Claude Partner Network.
> https://t.co/zE5DKPNhYM

[ @affine_io - SN120 ]
AFFINE-XXIX beat the Qwen3-32B baseline on SWE-Rebench, SWE-Multi, HumanEval and MCP-Agent benchmarks, while staying close on BBH.
> https://t.co/LSLFP1fuqU

[ @VantaTrading - SN8 ]
Vanta Trading crossed 2000 users after launching free $1k eval accounts and cutting prices by 55% across all challenges.
> https://t.co/O0Eqp3ekvY

[ @SwarmSubnet - SN124 ]
Swarm announced SOTApilot, an open-source AI drone autonomy model with 95.34% success on its UAV navigation benchmark.
> https://t.co/UnC4cGPsS0

[ @blockmachine_io - SN19 ]
Blockmachine launched Ethereum RPC.
> https://t.co/XyTFyf8pBf

[ @TrajectoryRL - SN11 ]
TrajectoryRL is expanding SN11’s skill competition from skill packs to miner-submitted finetuned models.
> https://t.co/cLxI0uKmy0

[ @heydittoai - SN118 ]
Ditto reached 1000 users.
> https://t.co/sLdkXHXdSZ

[ @theminos_ai - SN107 ]
Minos has run over 37,000 variant-calling evaluations on chromosome 21, with submissions improving by 10.21% on average.
> https://t.co/pqGUSFfxXQ

[ @minotaursubnet - SN112 ]
Minotaur launched its website and opened beta access to its DEX Aggregator.
> https://t.co/YO8qfkEuNH

[ @ReadyAI_ - SN33 ]
ReadyAI launched a revenue dashboard showing real-time demand for SN33’s structured data pipeline.
> https://t.co/6DB9XqVWdE

[ @say_gm_ - SN28 ]
Good Morning published the roadmap for its AI gateway running in a TEE, now live on testnet with mainnet beta next.
> https://t.co/AqePX5wEAA

[ @EndureNet - SN30 ]
Endure is integrating @SynthdataCo's forecasts into its DeFi risk engines.
> https://t.co/1xU2bpDdpF

[ @eirel_ai - SN36 ]
Eirel released its first product, offering deep research, image generation, web search and agent tools across multiple model families.
> https://t.co/XvD7V1uzIq

[ @adtao_ppcrebel - SN21 ]
@dsvfund took an OTC position in the SN21 alpha token.
> https://t.co/FwRoiRTOuA

SUBNET LAUNCH

[ @DeSciClaims - SN111 ]
Claims is launching as SN111 to build a claim-evidence graph that turns scientific literature into machine-readable data for AI reasoning.
> https://t.co/bTJIiR7JlE

PODCASTS & ARTICLES
@opentensor Novelty Search hosted by @const_reborn with @zipcodenetwork
> https://t.co/ueU7v8gnpy

@TAO_dot_com Episode 14 with @Carrot_____1 and @KeithSingery
> https://t.co/aFN9si9eU4

@gordonfrayne podcast with @josercaldera from Yanez
> https://t.co/6pXNdSHPkV

@gordonfrayne podcast with @knakamor from Vocence
> https://t.co/FKBYJ69YWv

@AltcoinMillie podcast with @MaxScore from Score
> https://t.co/fw2EymKeLO

@AltcoinMillie podcast with @zeussubnet
> https://t.co/j54g9qxzwa

@TAO_dot_com article “The Impact of Conviction”
> https://t.co/E3gjm7ZuG4

[Stev Reggy @ReggyStev]

Root validators should not have completely arbitrary freedom under Root Reborn.
Instead of “I give SN X weight 0.18,” they should score subnets on a predefined grid: product demand, PoUW quality, validator quality, miner diversity, revenue, transparency, tokenomics honesty, etc.
Mathematically: do not give one arbitrary score x.
Give a distribution x_i​ over predefined grades J_i​, with ∑x_i=1, and compute S=∑(x_i) (J_i).
Each score x_i should ideally require some evidence.
This reveals uncertainty and makes validator judgment auditable.
Then allocation weights should be derived from these scores, with caps for conflicts, liquidity, self-dealing, and concentration.
Root Reborn is powerful only if validators become serious subnet analysts, not market-cap copiers. Structured scoring is how you make that visible.

Validators should publish the scoring decomposition that produced it. For example, every subnet could be scored on fixed dimensions:
J_1 - External product demand (0–5)
J_2 - PoUW quality (0–5)
J_3 - Validator quality (0–5)
J_4 - Miner diversity (0–5)
J_5 - Product testability (0–5)
J_6 - Revenue/paid usage (0–5)
J_7 - Tokenomics honesty (0–5)
J_8 - Transparency/audits/dashboards (0–5)
J_9 - Security/operational reliability (0–5)

Each J_i is an integer from (0,5).

Then each root validator can still choose its philosophy, but within a structured system.

[xxxxxx @dc4348]

@LordRagnarao @TAOTemplar @markjeffrey Chute the only one worthy

[Chutes @chutes_ai]

The strongest open-source coding model is live on Chutes: GLM-5.2.

@Zai_org 's new flagship scores 81.0 on Terminal-Bench 2.1, within four points of Claude Opus 4.8 (85.0) and ahead of Gemini 3.1 Pro.

On SWE-bench Pro it hits 62.1, up from GLM-5.1's 58.4.

Native 1M-token context. A new attention design, IndexShare, cuts per-token compute by 2.9x at full context.

Runs as a TEE chute, so the GPU operators serving the model can't read your prompts or outputs. Confidential compute at the hardware level.

$1.40 in / $4.40 out per million tokens.

Run it now: https://t.co/mmdVEVFf3D

[NDubbs @TheNDubbs]

Big change possibly (hopefully) coming to bittensor:native

If you're invested in subnets, be sure to be invested in the top teams on Bittensor that have been here, killing it, for a while. They will be the biggest beneficiaries of the changes possibly coming.

Finally, I can hold the subnets I *want* to hold without auto-sell pressure. I feel like I've been forced to search for hidden alpha in "low-cap gems" scouring discords, X's and githubs for the past year.

Now, I can allocate to the QUALITY subnets and not worry about them bleeding constantly.

Root Reborn is GOOD for quality teams and subnets - not great for grifters.

[Shashii τao @DkingYooo18516]

Chutes just dropped another strong week of shipping.
@chutes_ai stayed heads-down instead of heading to Proof of Talk in Paris and the results are impressive.

Key updates from last week:

- Full TEE across the entire stack not just models, but the whole platform now runs in secure hardware enclaves.
- First Blackwell B200/B300 nodes live → bigger models, higher throughput.
- Validator fully rebuilt: autoscales, cheaper to run, tighter security.
Two new models dropped:
- Validator fully rebuilt: autoscales, cheaper to run, tighter security. hardware strong reasoning & agent performance, 256K context, runs fully in TEE.
- ACE-Step 1.5 XL high-quality music generation (lyrics → full track with vocals).
- New integrations: Now available via OpenGPU (Relay) and as a launch partner for Respan’s AI Gateway making decentralized, TEE-secured inference even easier for teams.

Chutes continues to push the bar on secure, decentralized inference while keeping the focus on real product delivery.
The team is clearly building for the long haul.

[Chutes @chutes_ai]

That's a wrap on the APU AIC AI Marathon.

It closed out on June 6, and we have been sitting with it since.

Teams pitched what they had spent the week building and the judges made their calls.

Jobest took the win for a full AI recruiter platform that did more than most expected from a few days of work.

But the projects were never the point. The room was.

Students who showed up unsure what decentralized AI even meant, and left having built something real on it.

That is why Chutes put its name on this.

The next wave of builders should not be gated behind cloud bills and closed models. Hand them open models and real compute, and they show you what they can do.

First of many to come, Malaysia's done.. what's next? 👀

https://t.co/guboAo3QjE

[taostats τ @taostats]

NEWS: @chutes_ai now on @openGPUnetwork making TEE-enabled models accessible to teams without wallets or infra setup.

Decentralized infra in action: pulling compute and models out of data centers and spreading them across the globe.

[Aphτ 🌊α @tao_Alph]

It’s time to have Chutes SN64 in a long term portfolio!!
@chutes_ai

[xxxxxx @dc4348]

@jon_durbin 🔥

[Jon Durbin @jon_durbin]

If your data pipeline can't take a giant open source vulkan/opencl game and turn it into a long horizon task that does thousands of steps and compactions/etc. that would take ~5 days at ~150tps what are you even doing?

[defimorphosis @DeFimorphosis]

take notes.

$TAO flowing into subnets it will only increase from here. It will be an organic, sustained uptrend.

Do not get lost in micro/mid-cap subnet speculation.

Spot decent-sized liquidity pools, great teams and great product. This is the subnet takeover

[Tyler DurdΞth @tylerdurdeth]

It was always Chutes. And always will be.

#1 subnet on Bittensor

[Tyler DurdΞth @tylerdurdeth]

Best explainer of the proposed new $TAO Root yield mechanics I have seen.

The more I think about it the more I like it: the bottom subnets are not worse off (their alpha gets sold via auto-sell right now anyway) while the best subnets will get advantaged.

Bittensor started suffering from a lack of price differentiation between top and bottom subnets, and this is a great way to amplify the gap.

Looking forward to reading more PoV but this feels overall positive for the ecosystem.

[Chutes @chutes_ai]

Week 5. Heads down and shipping.

Here's everything that went live:

→ Full TEE across the whole stack, first Blackwell B200/B300 nodes online, validator rebuilt
→ Nemotron 3 Ultra live: NVIDIA's 550B frontier open model, in TEE
→ ACE-Step music model live. Write the lyrics, get a finished track
→ Now a provider on OpenGPU, and a launch partner on Respan's AI Gateway
→ More Parallax explainers landing next week

Full breakdown below.

[Jesus Martinez @JesusMartinez]

https://t.co/UuwQHXVA7Z

[xxxxxx @dc4348]

@JesusMartinez So what subnets in “your mind” will benefit from this change?

[arkheτ.hl @arkhet]

@const_reborn chutes having their own validator is giga clutch

there’s 0 take, it’s all passed through to alpha stakers

[xxxxxx @dc4348]

@chutes_ai LFG