Mars retweeted
Case in point... THOU SHALT NOT COMPETE.
- AI/LLM development generally
- distributed training
- pretraining pipelines
- ML accelerator design
- cybersecurity
- chemistry
Already disclosed as nerfed for the plebs. What's next on the lobotomization list?
The dream state would be models within a few IQ points of the frontier labs (probably task specific first, depth before breadth), with a "training-as-a-service" arm to achieve breadth (inc. RL), all on TEE with true privacy, served at a fraction of the price of frontier models with significantly less hardware, thereby commoditizing and democratizing intelligence to the maximum extent possible.
The entire world's knowledge trains these models, they should in turn be trained by and accessible to the entire world.
The thing about privacy is, it's not really just about privacy. YOU are the product (builders) by ignoring privacy concerns.
People are building the massive golden trillion+ dollar moats for the various labs for them by giving free RL data constantly. It's great, for a few, not so great for everyone else. Need to make sure AI remains accessible and available to everyone, always.
If AI gets concentrated into a handful of players, it would mean those teams and maybe even a single person within ultimately gets to decide if you are allowed to multiply matrices in particular ways or not, not to mention constant surveillance etc.
Parallax aims to be the destroyer of moats.
Mars retweeted
Novelty Search tonight.
2 protocol upgrades to Subtensor.
Conviction + TAO Flow V2
2 subnet teams building on Bittensor.
\\ @cacheon_ai :: #SN14
\\ @Bitrecs :: #SN122
Live via Bittensor Discord
Hosted by @const_reborn
9PM UTC / 5PM EDT
Mars retweeted
Vidaio x Pip Studios — Joint Venture
We are beyond excited to announce this joint venture! This is bigger than a partnership. It’s a gateway to the global media ecosystem.
Through this joint venture, VidaioOS is now positioned directly within Pip Studios’ network spanning major studios, platforms, and content owners, including Netflix, Amazon, Sony, Universal, Paramount, and more.
This means:
→ Direct access to enterprise workflows
→ Real-world video workloads entering the network
→ Faster adoption at scale
Pip Studios is part of the TPN (Trusted Partner Network), the industry standard for secure content handling. That places VidaioOS inside trusted pipelines used by the world’s biggest players.
This isn’t about selling individual tools, its about embedding an AI-native video infrastructure layer across an existing global client base. Think of this as a gateway to 50+ major partnerships.
We’re not just building the future of video infrastructure; we’re now connected to where it already operates.
Mars retweeted
Learn about our locked stake revolution : CONVICTION
Live with @const_reborn
Wednesday 7pm UTC in the Bittensor discord: discord.com/invite/bittensor
Mars retweeted
1) what
Introducing GLM-5.1: The Next Level of Open Source
- Top-Tier Performance: #1 in open source and #3 globally across SWE-Bench Pro, Terminal-Bench, and NL2Repo.
- Built for Long-Horizon Tasks: Runs autonomously for 8 hours, refining strategies through thousands of iterations.
Blog: z.ai/blog/glm-5.1
Weights: huggingface.co/zai-org/GLM-5…
API: docs.z.ai/guides/llm/glm-5.1
Coding Plan: z.ai/subscribe
Coming to chat.z.ai in the next few days.
Mars retweeted
Bittensor Ecosystem Highlights of the Week #55
// SUBNET UPDATES & ACHIEVEMENTS
➤ @404gen_ SN17
404 unveiled Atlas, its application layer for enterprises.
(bit.ly/4siJNpU)
➤ @TargonCompute SN4
They announced TargonOS, an OS built on top of their TVM that will allow secure ingress and egress of consumer-grade GPUs onto the Targon network.
(bit.ly/4spxaJO)
➤ @chutes_ai SN64
OpenRouter updated its provider status, and Chutes is now in its default routing.
(bit.ly/3PQdCAD)
➤ @ReadyAI_ SN33
They launched their new website, and their llms.txt MCP is now live.
(xitter.cf/DavFields/status/20386…)
➤ @hippius_subnet SN75
Hippius Desktop App v2 is out, with a new sync engine built on Arion.
(xitter.cf/hippius_subnet/status/…)
➤ @tplr_ai SN3
Miners on Templar’s Crusades pushed a 7B model on 4xA100s from 66% MFU to 74.5% MFU.
(xitter.cf/tplr_ai/status/2039691…)
➤ @b1m_ai SN105
They managed to move 500MB in 21 to 29 seconds consistently with just 20 to 30 workers, showing Beam’s distributed bandwidth network can operate at the level of traditional transfer providers.
(xitter.cf/b1m_ai/status/20397197…)
➤ @webuildscore SN44
They hit 95.8% accuracy on the person detection benchmark.
(xitter.cf/MaxScore/status/203905…)
➤ @NiomeAI SN55
They partnered with Flore, a personalized microbiome health solution.
(xitter.cf/NiomeAI/status/2039042…)
➤ @numinous_ai SN6
@crunchDAO’s ML engineer network can now mine on Numinous.
(xitter.cf/numinous_ai/status/203…)
➤ @arbos_born (Distil) SN97
Within 24 hours, miners pushed past Qwen’s own distilled model and beat it on standardized benchmarks.
(xitter.cf/arbos_born/status/2038…)
➤ @resilabsai SN46
RESI reached 90% remote appraisal accuracy in just 2 months.
(xitter.cf/resilabsai/status/2040…)
➤ @SwarmSubnet SN124
They unveiled SOTApilot, the world’s largest benchmark for autonomous drone flight.
(xitter.cf/SwarmSubnet/status/203…)
➤ @HermesSubnet SN82
They released their OpenClaw-compatible skill.
(xitter.cf/HermesSubnet/status/20…)
➤ @Loosh_ai SN78
Loosh shared the changes they’re planning to roll out on the subnet.
(xitter.cf/Loosh_ai/status/203980…)
➤ @babelbit SN59
Babelbit is updating its incentive mechanism.
(xitter.cf/babelbit/status/203971…)
➤ @theminos_ai SN107
Minos announced a strategic partnership with @manifoldlabs.
(xitter.cf/theminos_ai/status/203…)
// SUBNET LAUNCHES
➤ @nexisgen_ai SN70
NexisGen is a subnet built by @rendix_network that aims to deliver on-chain verified datasets.
(xitter.cf/nexisgen_ai/status/203…)
// BITTENSOR ECOSYSTEM
➤ @bt_commons
Over 100 people from the Bittensor community came together for Breakout in San Francisco.
(xitter.cf/bt_commons/status/2039…)
➤ @btlabs_ai x @crunchDAO
They partnered to improve the quality of Bittensor mining.
(xitter.cf/btlabs_ai/status/20387…)
➤ @TAOInstitute_
They’re teasing “The Bloomberg Terminal for Bittensor.”
(xitter.cf/TAOInstitute_/status/2…)
➤ @YumaGroup
Yuma partnered with @ColossusItaly to let institutions delegate TAO to Yuma’s validator directly from custody.
(xitter.cf/YumaGroup/status/20390…)
@wearetalisman is now supported by Yuma’s validator.
(xitter.cf/YumaGroup/status/20386…)
// PODCASTS & ARTICLES
➤ Hash Rate 163 by @markjeffrey with @Nic_B1m from Beam.
(xitter.cf/markjeffrey/status/204…)
➤ @jollygreenmoney podcast with @MaxScore from Score.
(xitter.cf/jollygreenmoney/status…)
➤ @taodotcom Episode 11 with @Carrot_____1 and @KeithSingery
(xitter.cf/TAO_dot_com/status/204…)
➤ @JesusMartinez podcast with @DistStateAndMe from Covenant
(xitter.cf/JesusMartinez/status/2…)
➤ @supercyclepod by @haitzu with Wei
(xitter.cf/haitzu/status/20397127…)
➤ Yuma podcast with Hermes
(xitter.cf/YumaGroup/status/20397…)
➤ @mogmachine article “State of Bittensor: How We Capitalize and Become a Household Name”
(xitter.cf/mogmachine/status/2038…)
$TAO
Mars retweeted
Decentralized RL as fast as centralized RL for LLMs.
Bittensor SN81 @grail_ai has shattered the bandwidth barrier with PULSE (Patch Updates via Lossless Sparse Encoding).
"By identifying the 99% weight sparsity inherent in Adam-bounded updates, we've achieved a 100x reduction in weight synchronization - dropping 14GB transfers to just ~140MB "
@erfan_mhi + @const_reborn :: @covenant_ai
Mars retweeted
tao has something for everyone
it appeals to left and right curve
people within crypto and outside
you can focus on the tech or the social layer
rare to find projects like that
Mars retweeted
Bittensor Ecosystem Highlights of the Week #53
// SUBNET UPDATES & ACHIEVEMENTS
➤ @tplr_ai SN3
Covenant-72B was discussed by @chamath with @nvidia CEO Jensen Huang.
(bit.ly/3PnyjUe)
Crusades, Templar’s live training optimization competition, hit 63% MFU on 2×A100s, a new high for getting more real training work out of the same GPUs.
(bit.ly/4smw2an)
➤ @LeadpoetAI SN71
Leadpoet released its product launch presentation, introducing a service designed to deliver ready-to-buy prospects on demand.
(bit.ly/4t3VK3y)
➤ @Data_SN13 SN13
They introduced ‘dv’, a Rust CLI for querying real-time social data from X and Reddit.
(xitter.cf/Data_SN13/status/20346…)
➤ @webuildscore SN44
Score partnered with @Eyeballfootball, a video scouting and analysis platform, to bring Vision AI to youth football.
(xitter.cf/webuildscore/status/20…)
@stillcorecap invested in Score’s alpha token.
(xitter.cf/markjeffrey/status/203…)
➤ @chutes_ai SN64
They teamed up with @apuaic for a workshop, and they’re also teasing a big hackathon with APU AIC.
(xitter.cf/chutes_ai/status/20335…)
They also dropped a community note highlighting how their revenue per token and per GPU increased despite lower volume.
(xitter.cf/chutes_ai/status/20351…)
➤ @Bitcast_network SN93
Bitcast unveiled @Stitch3_ai, its new dedicated X platform.
(xitter.cf/Bitcast_network/status…)
Stitch3 also partnered with @LunarStrategy, a Web3 marketing agency, to onboard creators onto its platform.
(xitter.cf/Stitch3_ai/status/2034…)
➤ @vidaio_ SN85
They published a launch video teasing VidaioOS.
(xitter.cf/vidaio_/status/2034680…)
➤ @bitmind SN34
Bitmind dropped DFResearch, its deepfake research training toolkit inspired by AutoResearch and Arbos.
(xitter.cf/kenjon/status/20340112…)
They also released Mind Security, their OpenClaw skill.
(xitter.cf/bitmind/status/2034300…)
➤ @grail_ai SN81
They used AutoResearch to make Grail’s GRPO trainer 1.8x faster on a single B200.
(xitter.cf/grail_ai/status/203438…)
➤ @Apex_SN1 SN1
They built a simulation of the @IOTA_SN9 communication network on Apex, designed as a testing ground for experiments and for developing novel algorithms to improve the speed and quality of model training.
(xitter.cf/Apex_SN1/status/203466…)
➤ @metanova_labs SN68
ArboNOVA, Metanova’s system for linking molecules to relevant patents, reached an 85.4% best hit rate using only open data and tools.
(xitter.cf/metanova_labs/status/2…)
➤ @vocence_bt SN102
Their first product and dashboard are now live.
(xitter.cf/vocence_bt/status/2033…)
➤ @TargonCompute SN4
Targon has been accepted into Nvidia Inception for startups.
(xitter.cf/TargonCompute/status/2…)
➤ @resilabsai SN46 x Chutes SN64
Resi is now running inference through Chutes’ infrastructure.
(xitter.cf/chutes_ai/status/20339…)
➤ @vocence_bt SN102 x Chutes SN64
Vocence is now deploying PromptTTS models as private chutes on Chutes’ infrastructure.
(xitter.cf/chutes_ai/status/20342…)
➤ @ai_detection SN32 x @LeadpoetAI SN71
It’s AI partnered with Leadpoet to find qualified leads for its solution.
(xitter.cf/ai_detection/status/20…)
➤ @theminos_ai SN107
Their GitHub repo is now live.
(xitter.cf/theminos_ai/status/203…)
➤ @yanez__ai SN54
Yanez partnered with NiminoData, a leading compliance and risk data provider with a global client base.
(xitter.cf/yanez__ai/status/20346…)
They also showcased their progress in multimodal identity verification, the evolution of their incentive mechanism, and their achievements.
(xitter.cf/yanez__ai/status/20343…)
➤ @SwarmSubnet SN124
They secured a strategic investment from @Ever_BlueGreen, with 1000 TAO committed to the pool.
(xitter.cf/SwarmSubnet/status/203…)
➤ @b1m_ai SN105
Beam dropped an article explaining their architecture, network design, and how everything works.
(xitter.cf/b1m_ai/status/20350512…)
➤ @HermesSubnet SN82
Hermes now has new branding and a new website to better distinguish itself from @SubQueryNetwork.
(xitter.cf/HermesSubnet/status/20…)
// BITTENSOR ECOSYSTEM
➤ @SafelloAB
Safello’s staked $TAO ETP is now live on Nasdaq Stockholm.
(xitter.cf/FrankSchuil/status/203…)
// PODCASTS & ARTICLES
➤ @opentensor Novelty Search with @const_reborn on Arbos
(xitter.cf/opentensor/status/2034…)
➤ @supercyclepod with @markjeffrey and @rob_svrn from @stillcorecap
(xitter.cf/supercyclepod/status/2…)
➤ @VenturaLabs podcast Ep. 86 with @mega_tao
(xitter.cf/VenturaLabs/status/203…)
➤ My article “The Ultimate X Playbook for Subnet Owners”
(xitter.cf/Victor_crypto_2/status…)
$TAO
Mars retweeted
On the @theallinpod this week, @chamath asked @nvidia CEO Jensen Huang about decentralized AI training, calling our Covenant-72B run "a pretty crazy technical accomplishment."
One correction: it's 72 billion parameters, not four. Trained permissionlessly across 70+ contributors on commodity internet. The largest model ever pre-trained on fully decentralized infrastructure.
Jensen's answer is worth hearing too.
Mars retweeted
We just completed the largest decentralised LLM pre-training run in history: Covenant-72B. Permissionless, on Bittensor subnet 3.
72B parameters. ~1.1T tokens. Commodity internet. No centralized cluster. No whitelist. Anyone with GPUs could join or leave freely.
1/n
Mars retweeted
Prometheus stole fire from the gods and gave it to man.
We're doing the same with frontier intelligence.
Covenant is a permissionless AI lab ; anyone can build, train, and run intelligence without asking permission.
@tplr_ai pretrains. @grail_ai post-trains. @basilic_ai is the compute beneath it all.
The fire is yours now.
Mars retweeted
BITTENSOR ON-SITE HACKATHON
San Francisco :: @frontiertower :: 995 Market St
On-site subnet sprint: Mar 11, 9:00 AM → Mar 15, 1:00 PM
Build a sovereign AI layer as a Bittensor subnet ::
>> Applications open ↓
Judging criteria for your subnet build:
- Market demand // is this worth building?
- Miner task // what’s the work?
- Validator scoring // how is quality measured?
- Incentive design // do the economics drive improvements?
- Economic viability // can this sustain itself?
- Technical feasibility // can this actually be built and launched?
Prizes (Top 3 teams):
- Up to 1000 TAO discretionary funding + Investor pitch :: @UnsupervisedCap
- 50 TAO toward mainnet registration :: @CrucibleLabs
- $5,000 GPU compute credits :: @basilic_ai
- $2000 :: Cash bounty: $1000, $600, $400
- Launch pathway :: post-event support
> Judges :: @bitmind @DistStateAndMe @shibshib89 @6329
> Partners :: @FundingCommons @protocollabs @frontiertower @RedwoodNorthTAO
Mars retweeted
🇨🇳 @const_reborn at Tsinghua university, Beijing
(Full presentation) // @HackQuest_
Mars retweeted
Reminder: Keep building.
Think it. Pitch it. Build it. τ
Bittensor Subnet Ideathon (Round 1) closes Feb 25.
- $18K cash prize pool
- Up to 𝞃1000 discretionary funding from @UnsupervisedCap
- $5K @basilic_ai compute credits
- Direct entry to @bitstarterai accelerator
Registration live → build the next subnet on Bittensor.
hackquest.io/hackathons/Bitt…
Own the future.
#Bittensor $TAO #SubnetIdeathon @hackquest_
Mars retweeted
Build on Bittensor: IIT Delhi Workshop 🇮🇳
We’re bringing builder education directly to campus:
• Intro to @opentensor subnet design
• Brainstorming for our Bittensor Subnet Ideathon ($18K in prizes 💰)
• Live feedback from Nick, Affine core contributor
Jan 18 | RSVP 👇
luma.com/1a1je6io
Mars retweeted
✨ New #MEXC Kickstarter ✨
🔹 #SN75 @hippius_subnet
🔹 Trading: Oct 24, 2025, 11:00 (UTC)
Read More 👉 mexc.com/announcements/artic…
