Partner at Bittensor Fund @stillcorecap Hash Rate podcast
Los Angeles
Joined December 2007
- Tweets98.1K
- Following2K
- Followers74K
- Likes106K
Pinned Tweet
Hash Rate - Ep. 189: Jev on Bittensor SN92
๐ง Guest: @airesearch12 of @decisionmodels_ and @reaxlabs
Rent Jev models: decisionmodels.io
Mine SN92: reax.co
00:00 Introduction
02:02 Viral success of Florian's subnet announcement
03:26 Launch of decisionmodels.io
04:42 What's Jev?
06:16 Jev models v. traditional LLMs
08:26 Speed, cost, and intelligence
13:06 Jevbench
15:10 Open source Jev
22:07 Jev models in $TAO Subnets
27:24 Mining
33:24 Leaderboard
41:04 Decentralized Training of Jev models
49:37 Impact on traditional AI giants
55:19 The democratization of AI
Coming up in 15 minutes:
๐จ Subnet Summer TAO X Spaces - October 28, 2026 | Wednesday at 5PM GMT
๐ง๐ผ๐ฝ๐ถ๐ฐ: ๐ง๐ต๐ฒ ๐๐ถ๐ฟ๐๐ ๐ฃ๐ผ๐๐-๐๐ฎ๐น๐๐ถ๐ป๐ด ๐ฌ๐ฒ๐ฎ๐ฟ: ๐ช๐ต๐ฎ๐ ๐๐ ๐ฅ๐ฒ๐๐ฒ๐ฎ๐น๐ฒ๐ฑ ๐๐ฏ๐ผ๐๐ ๐ง๐๐ขโ๐ ๐ฃ๐ฎ๐๐ต ๐๐ผ ๐ฎ๐ฌ๐ฎ๐ณ
The halving didnโt just cut emissions. It stress-tested every part of Bittensor at once.
Four voices, four angles: capital, on-chain data, building, and validating.
Speakers:
โข Mark Jeffrey (@markjeffrey) - @stillcorecap
โข Mog (@mogmachine) - @taostats | @hippius_cloud | @vidaio_ | @blockmachine_io | @say_gm_
โข Max Sebti (@MaxSebti) - @webuildscore (SN44)
โข Tribit (@iamtheloraxnow) - Medulla Labs | @subnetsummerTAO
โข Bryden (@SubnetSummerT) - @subnetsummerTAO
Join us this Wednesday for a high-signal conversation on what the first post-halving year means for Bittensorโs path to 2027.
Set your reminder and join the discussion.
Mark Jeffrey retweeted
Chutes is now live on @say_gm_
SayGM gives you one API key for Claude, GPT, Gemini and open-source models.
When you pick one of SayGMโs private models, it can now run on Chutes, inside secure hardware where no one can read your prompts. Not even us.
See the full article here: chutes.ai/news/chutes-modelsโฆ
Nothing in the current rash of AI math papers suggests that what Elon says here has changed AT ALL.
Yes. AI is breaking SOME math VERY fast.
But not the KEY math used for crypto (and all encryption -- AI login, email login, health login, credit cards, banks, etc.)
Robot magicians now loading ....
(1/5) 225 consecutive Baoding ball rotations. One less finger. Zero cameras. Over 5 minutes without a drop.
We humans can manipulate objects in our hands with our eyes closed, relying on touch and a sense of how our fingers move. Can a robot hand do the same, with two balls that keep colliding with each other?
We trained a four-fingered Shadow Hand Lite in simulation to perform blind Baoding ball rotation, then deployed it on real hardware with no fine-tuning. We provide no vision, no depth and no motion capture. Just proprioception and 16 binary touch signals.
Paper: arxiv.org/abs/2610.23438
Website: noeyesnoproblem.github.io
Mark Jeffrey retweeted
Bittensor allows us to turn physics model development into an industry ready product.
How do we define that?
We don't. The industry already does.
The US Military has a standard for documenting why an AI model is suitable for a specific use case:
MIL-STD-3022
American Society of Mechanical Engineers is currently writing the first standard for verifying machine-learning models in engineering.
ASME V&V 70
We designed Carbon to meet this gold standard of AI model adoption in the engineering industry.
Bittensor. Required.
Why?
Miners: Auditable model designs
Validators: Adversarial/Independent V&V mechanism that can generate all the data needed to meet these standards.
Carbon will be the standard for deploying physics AI where being wrong isn't an option.
Much more on this to come...
$TAO
$TAO on Base and Robinhood ...
$0 to $3.3M worth of $TAO in less than 80 days ๐
That's how much TAO has been bridged to @base and @RobinhoodCrypto through ForeverMoney since we launched. Our bridge moves native TAO 1:1, and it's secured by @chainlink CCIP.
- Total bridged: 13,393.6 TAO (~$3.32M)
- Bridged to Base: 8,521.78 TAO (~$2.11M)
- Bridged to Robinhood Chain: 4,871.82 TAO (~$1.21M)
Wherever TAO goes next, it gets there through ForeverMoney. Link to our bridge below ๐
Mark Jeffrey retweeted
Thank you President Trump for helping the Bolsonaro family avoid political persecution by Lula and Moraes
Brazilian democracy faces an existential crisis, and the Bolsonaro family is bearing the brunt of it
Mark Jeffrey retweeted
I am long $TAO and love bittensor's focus on scaling AI
This is what a REAL 'asylum seeker' migrant looks like.
๐จ๐ง๐ท๐บ๐ธ Trump's administration is opening the door to Flรกvio Bolsonaro and his family after his election defeat in Brazil.
The U.S. State Department says it will grant them VISAS, just as Bolsonaro faces a court order to surrender his passports.
He reportedly left Brazil for the United States after losing to Lula, despite the order.
Source: AP / Writer: Oliver
The Bol Who Lived
I can confirm that @FlavioBolsonaro is no longer in Brazil and is on his way to the United States.
A US protective operation is underway to ensure his safety.
Revenue is flowing into Bittensor subnets.
And they're using it (in part) to buy n burn their subnet tokens.
Nearly $2M from Targon SN4 just now:
We have just repurchased and burned 1,772 TAO (~$445K) of SN4 alpha using organic revenues generated via Targon.com
We have also burned an additional 6,678 TAO (~$1.7M) worth of SN4 alpha, which also represents revenue generated via Targon
Subnet 32 detects whether text is AI or not -- now across several languages.
powered by Bittensor
Our multilingual accuracy results are in ๐ฆ๐ช๐ฉ๐ช๐ซ๐ท๐ช๐ธ
We benchmarked Itโs AI on DetectRL-X (ACL 2026), a new multilingual benchmark covering 6 real-world writing domains, modern LLMs and AI-edited text. ~12K human and AI texts per language:
๐ฆ๐ช Arabic: 99.8%
๐ฉ๐ช German: 99.5%
๐ซ๐ท French: 99.4%
๐ช๐ธ Spanish: 96.0%
But accuracy is only half the story. A detector that flags honest work is useless, so we also tested every language on real student writing:
๐ฆ๐ช 0.89% false positives (AR-AES)
๐ฉ๐ช 0.48% (DARIUS)
๐ซ๐ท 0.56% (E-CALM)
๐ช๐ธ 0.72% (CAES, 20K+ texts)
Under 1% in every language.
In English: 98.3% on RAID and #1 on MGTD, ahead of GPTZero and Originality AI.
Every benchmark, paper and dataset on one page๐
I am long $TAO and love bittensor's focus on scaling AI
RT @SubnetSummerTAO: ๐จ Subnet Summer TAO X Spaces - October 28, 2026 | Wednesday at 5 PM BST
๐ง๐ผ๐ฝ๐ถ๐ฐ: ๐ง๐ต๐ฒ ๐๐ถ๐ฟ๐๐ ๐ฃ๐ผ๐๐-๐๐ฎ๐น๐๐ถ๐ป๐ด ๐ฌ๐ฒ๐ฎ๐ฟ: ๐ช๐ต๐ฎ๐ ๐๐ ๐ฅโฆ
