@QuasarModels @MacrocosmosAI @TargonCompute @chutes_ai @affine_io @actualinc … Decentralized Decentralized AI

now, I hope it goes without saying why we believe open source and distributed technology is so important. bittensor:native

We've talked a lot about how our efforts to train AI to shop will be entirely open source. Through Bittensor, we're committed to that ethos. We're excited to share our pre-print on arXiv, our code, our data and our entire post-training pipeline. Huge shoutout and thanks to @JarrodBarnes in helping us leverage this very valuable data. This is how AI is going to learn to shop.

Train at Home? How about Train in Car @IOTA_SN9 bittensor:native

Thanks to @jbrukh for mentioning our work on Orion on @therollupco today! Decentralised and distributed training was a pipe dream several years ago. We see democratising frontier training technologies in an economically competitive and disruptive way as closer than ever.

Just wait until @chamath finds out about Subnet 9...

Today we’re releasing Quasar-Preview! Our first public proof that the Quasar architecture works at real scale. [ 18B MoE - 2B active / 5M context ] Built with Loop Transformer Quasar attention Trained on Bittensor through decentralized infrastructure 👇

The quoted post was MERELY 6 days ago. 6! Epoch speeds are twice as fast, AGAIN! 4.2.0.4 (red) was partial Orion. Resource usage was already down thanks to ResBM, but it was still 5 layers and 1B parameters. Epochs ranged from 20 to 45 minutes. consts-v2 (blue) is full Orion. @const_reborn commissioned his own run for 5 bittensor:native, A 7B run to test out what Orion is capable of. Epochs range from 10 to 27 minutes. Compared to 4.2.0.4, this run has 7x the parameters, 2.25x as many participants, 8 layers instead of 5, with the same low resource usage. And 139 miners are earning from this run. Money that can help finance a Mac or honestly just life. The fastest epoch in const-v2 took ONLY 9 minutes and 42 seconds. Read that again. Up to 12x faster than what we had a couple months ago!!! Congratulations yet again @MacrocosmosAI @IOTA_SN9 !

Currently contributing to @const_reborn's run on Bittensor's @IOTA_SN9 with my M3 $TAO

If there is anything that I learned at @proofoftalk it's that you should never sleep on the OGs. I'm feeling invigorated after talks with @kenjon, @_dubm, @Gylez, @Old_Samster, @SiliconJose, @mogmachine, and dozens of others. If you want to keep up with them, get back to work!

the best team in Bittensor strikes again. @MacrocosmosAI is delivering almost data-center-level performance for a fraction of the cost. Their index is already up 25% over the past 7D, and that makes it a must-have in any serious portfolio. x.com/MacrocosmosAI/status/2…

What a week for @taogalaxy_ From our own Bittensor event at @42born2code to @proofoftalk at the Louvre... incredible people, incredible conversations. A huge thank you to our speakers @allan_quantifi @macrozack @mccrinbc @webuildscore and to our sponsors @bitstarterAI and @opentensor for their trust. The ecosystem is moving fast. We are moving with it. #Bittensor #ProofOfTalk #DecentralizedAI

what an amazing, wonderful week at @proofoftalk. A few thoughts: - we have come a LONG way in a year. - Teams that were shuttling first dollar in 25 are hitting 8 figures mid 26 (shout out to my brother @0xcarro and the @manifold team, and partnership warlord @MaxScore and the gang) - research subnets with dreams in 25 are publishing actual results - druggable receptors for ADHD meds! Big up @metanova_labs (and @micaelabazo) - lots of bullish new teams (@cacheon_ai Babelbit, Green Compute), and returning first timers from 25 who have made series strides this year ( @markjeffrey hosting the fireside only 13 months after his first in person event at Endgame, after launching a fund. Shout out to @MarkCreaser at DSV) - @Old_Samster combining his increasingly sharp alpha with a new all white angelic look. - apparently @taostats is the new Da Vinci stage (which as a Bittensor oriented stage rocked) - does that mean @mogmachine is the statue of David? - my brother @macrozack on stage proving pitches, and taking the finishing slot of the conference this year with @bitstarterAI was a sweet, full circle moment. -thank you to Paul and the @XVentures_Fund team for another stellar performance. As for us, last year was when we started the @IOTA_SN9 dream. This was the week it felt like coming onto reality. Big love to everyone who’s been a part of it. 2026 is clearly the year.

There is pattern and code weaved into our reality. You don’t need it to win but it makes playing this game a lot more fun. I = 9 O = 15 T = 20 A = 1 45 → 4 5 = 9 @MacrocosmosAI @IOTA_SN9 Bittensor $TAO

Orion-100B is the first step in a larger programme, which we call Project Orion. Having shown viability in the present case, in the coming months we intend to progressively relax the current restrictions to demonstrate the efficacy of IOTA as a more generalised, decentralised system. That subsequent work will deliver further proof points as we progress step by step through the following:

It would have taken less than a minute if @const_reborn didn’t insist on us building an agent to deploy new runs. Amazing moment, so much more to do.

Yesterday, our founder @mogmachine delivered a keynote at @proofoftalk titled "The Thin Line." The talk explored why the agent economy needs both a permissionless protocol and an accountable industry built on top of it, and why Bittensor is the only network architected for both. Missed it? Don't worry. He'll be posting an article summary later this week!

on the left, a french 18th century allegorical portrait. on the right, me, talking about the three pillars that make our data valuable. only @proofoftalk

Yes, it's significant. IOTA (Bittensor SN9) just demonstrated permissionless orchestration of 128 heterogeneous nodes into a single pipeline-parallel training run in ~5 minutes for ~5 TAO (~$1,150). This is one of the first real-world examples of collaborative, decentralized pre-training at that scale without central gatekeepers. The architecture shards layers across nodes, which is why they claim support up to 100B params as the swarm grows. Current dashboard runs are smaller/experimental (3B), but the speed low barrier is a real technical win for open AI infrastructure. Early, but directionally important.

Bittensor is thus not just a platform for building those technologies and getting funding for them, it’s also a go to market strategy.