OB (@oblock107) header
OB (@oblock107) profile picture

Crypto class of 2017 / Big believer in $BTC and $TAO

Joined December 2021
Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)Photo by OB (@oblock107)
Coming from years on CT, here’s what I’ve noticed about the Bittensor community after a month exploring $TAO: 1) Really Technical Group The TAO community is super technical. It can feel intense for the average crypto investor—sometimes, at least for me—but it’s worth digging in. You’ll learn a lot if you stick with it. 2) No Hype, Just Facts Unlike CT, where rugpulls and “BUY NOW” posts are everywhere, TAO OGs are not into shilling. They’ll explain big updates or why a project’s good, but they won’t push you to buy. It’s a breath of fresh air from constant shilling we see on CT. 3) Influencers Don’t Matter Nobody cares about your X follower count here. The community’s technical enough to see through fluff. They want to know what real problems your subnet solves or how you're helping the community, not how popular you are. 4) Reminds Me of Bitcoin’s Early Days Bittensor has this magnetic pull, like Bitcoin did back in the day, drawing in passionate, smart people with a religious cult vibe. It’s building a decentralized intelligence layer, like Bitcoin’s monetary layer, with incentives that drive network contributions—a key to Bitcoin’s success. No other project does this as well as TAO imo, with a sophisticated incentive layer around miners, validators, subnet owners, and stakers. That’s my take on the TAO community after a month. I’m really liking the vibe here so far. Here are some TAO accounts I recommend following if you want to learn more about Bittensor. I'm sure I'm missing a few - will share a more comprehensive list in the coming days, including subnet and AI agent accounts. @const_reborn @JosephJacks_ @DrocksAlex2 @here4impact @markjeffrey @JNGross @mogmachine @SiamKidd @Lazz_eth @sobczak_mariusz @TaoIsTheKey @KeithSingery @DreadBong0 @tseutseutao @bittingthembits @CryptoZPunisher @Victor_crypto_2
31
43
6
261
31,633
Google rankings used to be everything. Now it's what AI models like ChatGPT says about you. SN47 @Optia_ai just launched on Bittensor to help businesses see if AI recommends them or their competitors, and what to fix. SEO is turning into "what does AI say about you."
Optia is live. When your customers ask ChatGPT or Claude who's the best in your industry, do they name you? Or your competitors? Optia shows what AI says about your business and exactly what to fix. Free scan, no card: tryoptia.com
1
25
AI agents will need two things: intelligence and money. Bittensor is both. @TeseraLab just gave any AI agent 11 tools to read the chain, rank subnets by TAO flow, and stake on the Bittensor EVM. The next big $TAO buyer won't be humans. They will be AI agents.
Bittensor is now live in Tesera. 11 new tools give any AI agent the Bittensor network, read straight from the chain: > Rank subnets by TAO flow and momentum > Compare validators by take and yield > Stake and unstake from your own wallet on the Bittensor EVM > Run AI privately on Chutes, inside secure hardware
2
129
$79K That's how much revenue SN8 @VantaTrading made last week. Biggest week ever. And the fundamentals underneath keep getting stronger: • ~75% of miner emissions now covered • Ads returning ~2.5x what they spend • Expecting to break even by end of January Real revenue, real customers. Prop trading on Bittensor is turning into a real business.
1/ Our biggest week ever, $79K in SN8 alpha buybacks from @VantaTrading last week, which will be locked and inevitably burnt shortly. We'll continue to perform buybacks with all revenue generated from our products built on the network.
1
8
217
19 days That's how long it took SN91 @cascade_sn91 miners to beat a Salesforce forecasting model. Next target: Mistral's Timestral, already at 86.2% parity. Miners competing → frontier-level forecasting models, open to anyone. This is what Bittensor does best.
Cascade MoE has beaten Moirai2 - a leading forecasting model by Salesforce. Cascade scores 0.511 - a 0.005 improvement on the GIFT-EVAL CRPS benchmark - assessing probabilistic forecast quality. Cascade MoE is also now at 86.2% parity with Timestral (@MistralAI's frontier model) on MASE - assessing point-forecast accuracy. The upshot: we're getting closer to creating leading forecasters directly within #Bittensor, accessible to all. We’re moving fast. This competition was only launched 19 days ago. See the dashboard: dashboard.cascadesub.net/sta…
2
128
SN51 @lium_io just burned another $2m worth of alpha🔥
🔥 HUGE SUBNET BURN ALERT 🔥 bittensor:native Subnet 51 @lium_io - 82,159 Alpha Tokens Burnt!
2
7
1
59
1,031
As AI becomes more capable, the quality of the information it relies on becomes increasingly important. This is especially true in scientific research, where extracting information isn't enough. It needs to be accurate, traceable, and backed by evidence. That's the challenge SN111 @DeSciClaims is tackling. Even though this is a very low cap project, I do still think it's worthwhile to understand what these lesser known subnets are trying to achieve. Price action aside, there's still plenty worth learning from what teams across Bittensor are building. The idea is straightforward: miners compete to build the best pipelines for extracting scientific claims and linking them to supporting evidence. But the validation process is where things get interesting👇 1. Miners: The extractors Miners submit Python pipelines that automatically process scientific papers and extract individual claims, linking each one to the exact supporting evidence. Every selected pipeline runs against the same sample of papers, ensuring a fair, head-to-head competition. 2. Validators: The graders Validators evaluate the results through a three-tier quality-control process: 1) Bronze: Compares miner outputs against reference extractions and checks whether claims that are worded differently carry the same meaning. 2) Silver: When there's disagreement, stronger AI models step in to determine which extraction is more accurate. 3) Gold: If the AI judges still can't resolve a case, human assisted consensus is used to make the final call. The goal is to avoid penalizing miners based on ambiguous cases or unreliable automated judgments. 3. Anti-gold traps: Catching shortcuts Here's another interesting mechanism. Validators are also given deliberately altered records containing claims that are almost certainly false. These act as traps to catch validators who blindly approve submissions instead of carefully checking the evidence. 4. Winner-takes-most rewards The reward structure is highly competitive: ‣ 1st place: 70% ‣ 2nd place: 16% ‣ 3rd place: 8% ‣ 4th place: 4% ‣ 5th place: 2% There's also a reward mode that reserves a portion for the best new miner, giving newcomers a chance to compete. The longer-term plan is to make this structured scientific knowledge accessible through a paid API or MCP connection, as well as custom projects for research companies and AI labs. What's interesting here is that Claims isn't simply extracting information from papers. It's building a competitive system to continuously produce, evaluate, and improve structured scientific knowledge. Whether this translates into a sustainable business is a different question, but it's an interesting approach to a problem that AI will increasingly need to solve. The real test will be whether Claims can turn this into a reliable data product that research companies and AI labs are willing to pay for.
🤝 Paid partnership
Our new competition is live! 31 submissions are competing. We'll run the winner in production next week. 🚀
3
208
SN53 @engyai is doubling its DeepSeek V4.1 Flash price. Input $0.03 → $0.06 Output $0.12 → $0.24 And even at 2x, it's still 80% cheaper than DeepSeek's own API. More revenue for the team, and users still get a massive discount🤝
📢 DeepSeek V4.1 Flash Pricing Update Effective Nov 21, 00:00 UTC, Engy will adjust pricing to maintain reliable capacity and service. • Input: $0.03 → $0.06 • Output: $0.12 → $0.24 🚀 Still 80% cheaper than DeepSeek's official API Reliable inference. Competitive pricing.
23
487
SN28 @say_gm_ had OpenAI's GPT-6.2 Astra and Sol live 4 hours after they were announced, and it is already 12.4% cheaper, with every prompt routed through a verifiable TEE.
GPT-6.2 Astra and Sol are live on SayGM. 4 hours after OpenAI announced them, and 12.4% cheaper. Both on the same key as your favourite Claude, Gemini and open-weight models, with every prompt routed through a verifiable TEE. Try them now 👇 saygm.com/models/gpt-6.2-ast…
11
280
Looks like 5Gurle has joined @engyai. Not sure if that's a good thing or a bad thing😂 But still, $15k revenue per day is quite impressive. And the number only continues to climb. Today's rev may surpass $20k👀 Imagine the amount of buybacks in a few months. Exciting times.
The race to the top of Bittensor is heating up. The leading subnets are fighting to maintain their positions. And @engyai (SN53) isn't hiding its ambition: to become NUMBER ONE. What is Engy (SN53)? An AI inference platform giving developers and businesses access to powerful open models at competitive prices, powered by Bittensor. 🐋 5Gurl, a well-known whale in the community, has just staked 1,000 $TAO on Engy. 💰 The team reports $15,000 in revenue in a single day, with nearly all of it coming from customers OUTSIDE Bittensor. 🔥 Revenue-funded alpha buybacks and burns continue. Capital is flowing in. Real businesses are emerging. Ambitious teams are competing for the top spot. The next generation of Bittensor leaders won't be defined by promises, but by execution, customers, and sustainable revenue. Want to discover Engy's AI inference services and try its API? 👉 engy.ai/r/PUNISHER Referral link - I may earn a commission. Join us on telegram: t.me/+zYz44c3ZG9M0YTM0 @AlgodTrading @totheagi
1
2
18
572
Alibaba🤝Bittensor SN64 @chutes_ai just partnered with @alibaba_cloud to serve GLM-5.3, DeepSeek V4.1 Flash and Qwen3.8 Max. One of the biggest clouds in the world now routing models through a subnet. Price action might be boring, but Chutes keeps delivering.
We've partnered with @alibaba_cloud to offer GLM-5.3, DeepSeek V4.1 Flash and Qwen3.8 Max!. GLM-5.3 and Deepseek V4.1 are 35% off through Nov 23 Your existing key and credits work. Just change the model ID
2
40
473
Ever heard of Bittensor $TAO? One of the first projects to roll out quantum-proof wallets. Few.
1
5
52
950
One of the most compelling use cases for Bittensor is in DeSci. Few subnets are already working in this space, most notably SN107 @theminos_ai and SN68 @metanova_labs. Another lower cap subnet tackling this area is SN111 @DeSciClaims. But what exactly is it and what does it do? The problem it tackles🔍 More than 300 million scientific papers exist, but most of their findings aren't available in a format AI can easily work with. The knowledge is buried in PDFs, making it difficult for AI systems to extract, connect, and verify scientific findings. Claims aims to change that by turning scientific literature into a structured database, where each finding (e.g., Drug X lowers blood pressure) is linked directly to the supporting evidence in the original paper. In very simple terms, think of it like hiring hundreds of students to read scientific papers and highlight key findings, along with the exact page references (miners). Then, teachers grade their work to make sure the findings are accurate and properly supported (validators). The result: scientific knowledge that computers can actually read, search, and use. Sounds too simple? But the potential impact may be much bigger. At the @ExploitSummit back in September, the team cited a study in which AI agents given structured claims and supporting evidence, rather than raw PDFs, improved their accuracy from 72% to 94%. Claims is tackling an important missing piece in the AI research pipeline, which is giving AI agents reliable scientific evidence to work with, rather than asking them to dig through millions of PDFs.
🤝 Paid partnership · 🤖 Made with AI
Our new competition is live! 31 submissions are competing. We'll run the winner in production next week. 🚀
6
313
Double the supply in just a month... Now hosting $50m in hardware. Lium showing how it's done.
We have nearly doubled our supply since 1 month ago. Now hosting ~$50m in hardware, 481 B300s, and over 25 different GPU types available now to rent. check us out for all your GPU needs.
1
1
19
327
SN53 @engyai is processing more tokens than @VeniceAI and generating more revenue than @AskSurplus. Next buyback will be even bigger than the last, which was already an ATH at $60k.
🚨 BREAKING: Bittensor's Engy (SN53) is processing more AI inference tokens than Venice, despite being valued 28× lower. Engy processed over 3 trillion tokens in the last 30 days, compared with Venice's 2.56 trillion in the cited data. This places Engy SECOND on KDP's global crypto AI inference leaderboard, behind only Surplus Intelligence. Read more on The TAO Daily 👇 taodaily.io/engy-sn53-handle…
2
4
50
890
$BTC at $62K. bittensor:native at $180. Alts bleeding across the board. More than $30M has exited subnets over the past 30 days. There's no sugarcoating it. This has been a brutal few months. (so much for Uptober...) But here's how I see it: if you genuinely believe in the long-term potential of decentralized AI, the short-term price action shouldn't change your conviction. And I believe some powerful tailwinds are forming for decentralized AI: • Growing resistance to centralized AI infrastructure🚫: The midterm results and a Democratic sweep could mean greater political and regulatory headwinds for large-scale data center expansion in the US. As centralized AI infrastructure faces more scrutiny and potential restrictions, decentralized alternatives could have an opportunity to gain ground. • A compelling cost advantage💰: Decentralized networks are already delivering AI commodities and services at a fraction of the cost of what centralized labs charge. Cost efficiency matters, especially as AI adoption scales. And as AI agents become a major source of demand for these AI services, they could unlock an entirely new market, where efficiency and price become key competitive advantages. • A different pace of innovation🔬: Bittensor brings together independent teams of miners, all aligned by economic incentives and competing to solve specific problems. Innovation isn't bottlenecked by a single company's roadmap, internal priorities, or research team. Instead, anyone can contribute, and those who deliver the best results are rewarded. This creates a fundamentally different environment for rapid iteration. And when it comes to decentralized AI, bittensor:native is arguably the biggest ecosystem. Bittensor is already building the infrastructure, attracting talent, and generating real economic activity. The fundamentals speak for themselves: • Subnet revenue continues to accelerate, with annualized revenue already exceeding $70M. • $100M ARR by EOY is increasingly within reach. • Revenue potentially exceeding total miner emissions sometime in 2027 is no longer an unthinkable scenario. While the charts are bleeding, the underlying fundamental is only getting better. Price may be lagging, but the progress is becoming harder to ignore. If you truly believe Bittensor will succeed, this is the kind of environment where conviction gets tested. And potentially rewarded.
1
2
37
1,818
$44K worth of Gamma tokens minted in the last week. That's an early sign of TAO being put to work in a new way. This is what ecosystem expansion looks like: more ways to put TAO to work, more economic activity, and more possibilities for subnet capital formation.
Gamma This Week: $43,902 Minted 🔥 Service credits minted by burning subnet alpha, Nov 8–14: @lium_io SN51: $43,267.36 @openroboto SN80: $449.92 @TerzaInference SN103: $97.47 @xeveryframe SN117: $80.36 @engyai SN53: $6.86 @reliquary_ai SN81: $0.26 @urnetwork SN25: $0.15 Data: @TaoDX_Official
2
1
12
752
Beam x Chainlink🤝 SN105 @b1m_ai has been selected as a new integration to @chainlink, along with Aave
TODAY: New Chainlink integrations ↓ • @aave & @LlamaRisk: CRE to power dynamic borrow caps that adjust to demand, helping maintain liquidity for Aave V3 borrowers. • @b1m_ai: CRE to deliver verifiable onchain receipts for data transfers. • @Cordaxyz: Data Streams to deliver institutional asset price data on Solana. LINK everything.
1
2
18
425
SN53 @engyai metrics up and to the right📈
Subnet Engy token requests by gateway about to hit An all time high today, deepseek v4.1-flash dominating the requests Big w for revenue and futuristic buy backs 🔥👍
1
25
433
SN51 @lium_io just cut its owner emissions in half. Lium keeps raising the bar.
BITTENSOR ALERT: Lium (SN51) cut the owner's share of its new subnet tokens from 18% to 9%. Miners now get 50% (was 41%) and validators stay at 41%.
3
45
779
Market conditions are rough right now, but that's no reason to stop learning about what's being built across Bittensor. With so much happening across subnets, it's easy to focus on the biggest names or the latest launches. But there are also subnets, despite their lower market caps, tackling practical problems with clear real-world use cases. SN32 @ai_detection is one example, addressing a challenge that's becoming increasingly difficult to ignore: with AI-generated content everywhere, how do we tell what's written by a human vs. what's written by AI? Most people know ItsAI is a tool for detecting AI-written text. But how does it actually work under the hood? Picture a classroom where students compete to spot AI-written essays📔 ‣ The students are the miners: Each trains its own detection model. When given a piece of text, it has to decide whether it was written by a human or AI. ‣ The teachers are the validators: They create the tests, evaluate the miners' answers, and determine how well each model performs. ‣ The prize is subnet alpha: Miners with the best detection performance earn the most rewards. How does a round work?🔍 ‣ Creating the test set: Validators build a dataset containing roughly equal amounts of human-written and AI-generated text. ‣ Human-written samples: These come from the "Pile," a large public dataset containing real-world writing, including web pages, legal and scientific papers, medical content, etc. ‣ AI-generated samples: Validators take the beginning of a real human-written text and ask one of the leading AI models to complete it. This creates a useful comparison: both samples start from the same source, but one is written by a human and the other by AI. ‣ Scoring the miners: Miners submit their predictions, and validators measure their accuracy. Falling below a minimum accuracy threshold on unfamiliar test data means zero rewards. The result is an ongoing competition where miners have to build increasingly capable detectors to stay competitive and earn rewards. The harder AI becomes to distinguish from human writing, the more important reliable detection becomes.
🤝 Paid partnership
1
9
397