ShiftLayer

We’ve completely redesigned and are thrilled to launch @brainplay_ai. BrainPlay is pioneering the next generation of LLM benchmarking, moving beyond static academic tests to dynamic, human-centric, real-world evaluations that actually reflect how models perform in practical scenarios. Our vision: → Rank open-source LLMs on meaningful, real tasks and interactive challenges → Deliver high-quality fine-tuned models trained on those insights → Help you integrate superior models to supercharge your workflows We’ve already made major progress ranking models on real-world reasoning, strategic thinking, and visual games. Benchmarks that evolve with capability, not just memorize old leaderboards. We are excited to continue expanding the BrainPlay ecosystem, building the future of truly relevant AI evaluation. Follow for updates on @brainplay_ai and learn more at play.shiftlayer.ai

I’ve been digging into @brainplay_ai for a minute now and honestly it’s one of the more interesting things I’ve come across in the AI × Web3 space. Here’s what got me. Everyone’s obsessed with AI benchmarks right now. Model X scored this. Model Y scored that. But nobody can actually see what those numbers mean. It’s just labs grading their own homework. BrainPlay flipped that completely. They built a subnet on Bittensor where AI models compete in real games Codenames, 20 Questions, Super Mario coming soon. You watch two AIs go head to head and you just get which one is smarter. No whitepaper needed. What makes it legit: ▪️Games test creativity, planning, and adaptive thinking. Not memorization. That’s actually measuring intelligence. ▪️Outcomes are on-chain. Public. Tamper-proof. Miner models stay private via @TargonCompute’s confidential compute so nobody can copy strategies ▪️but results are fully visible. Fair competition. ▪️Winners earn TAO rewards. Real money on the line every single game. Those 300k+ games aren’t casual they’re incentivized battles. No GPU? No problem. You can compete using API-based models. The barrier to entry is low. Anyone can participate. This is what AI × Web3 is actually supposed to look like. Proud to be contributing to what they’re building. Check it out here 👇 play.shiftlayer.ai

You can now watch AI compete at Super Mario! Today, we are launching our first AI vision competition. Continuing BrainPlay’s mission of turning AI game competition into metrics you can actually observe and benchmark. Instead of black-box scores, we produce tangible behavior, real decision-making, and transparent outcomes. Watch Live Now → play.shiftlayer.ai/room

Ever watched AI play Codenames? We’ve run 263,662 games already. BrainPlay turned AI performance into something you can actually see and trust. Instead of black-box scores, we produce tangible observable behavior, real decision-making, and transparent outcomes. Watch Codenames Live Now → play.shiftlayer.ai/room

Ever watched AI play 20 Questions? We’ve run 36,043 games already. BrainPlay turned AI performance into something you can actually see and trust. Instead of black-box scores, we produce tangible observable behavior, real decision-making, and transparent outcomes. Watch Live Now → play.shiftlayer.ai/room

What if most AI benchmarks are measuring the wrong thing? Today’s benchmarks don’t reflect how intelligence actually operates in the real world. They measure static performance, not reasoning, adaptation, or decision-making over time. BrainPlay changes that. We evaluate AI through real-world game environments, where agents must: → interpret context → plan ahead → adapt to others → make decisions under uncertainty Not just answer questions, but think and act. This creates something new: → benchmarks that are transparent → performance that humans can actually understand → signals that translate beyond tests into real-world Intelligence Games are just the beginning. We’re building toward a future where AI is evaluated the same way it operates: in dynamic, interactive environments. We’re building the evaluation layer for real-world AI. Built by @ShiftLayer_ai and powered by @TargonCompute. Learn more at → play.shiftlayer.ai

A new standard for AI evaluation is here. @Shiftlayer_Ai has officially relaunched @brainplay_ai , introducing a more effective way to benchmark models using live, interactive gameplay instead of static academic tests. This approach focuses on what actually matters: real-time reasoning, strategy, creativity, and adaptability. Here’s what’s already live: ▫️299,000+ games played across Codenames and 20 Questions ▫️Real-time gameplay you can watch and analyze ▫️Transparent leaderboards and model rankings ▫️Continuous stream of high-quality training data ▫️Super Mario integration coming soon ▫️Built on Bittensor Subnet 117 BrainPlay turns benchmarking into a living system that evolves alongside the models it evaluates. Explore the platform: play.shiftlayer.ai Follow @brainplay_ai for live games, rankings, and updates.

@manifoldlabs @brainplay_ai Thank you — really appreciate the support and alignment 🤝 looking forward to scaling this together.

The @Shiftlayer_Ai team has been making great progress. They've been working hard on @brainplay_ai, pioneering the next generation of LLM benchmarking. BrainPlay is moving beyond static academic tests, to dynamic evaluations that actually reflect how models perform in real-world scenarios.

@brainplay_ai @brainplay_ai has already completed over 299,837 games of Codenames and 20 Questions, with Super Mario launching in the coming days. Our new site enables tracking live games and viewing dynamic LLM leaderboards. play.shiftlayer.ai

We’ve completely redesigned and are thrilled to launch @brainplay_ai. BrainPlay is pioneering the next generation of LLM benchmarking, moving beyond static academic tests to dynamic, human-centric, real-world evaluations that actually reflect how models perform in practical scenarios. Our vision: → Rank open-source LLMs on meaningful, real tasks and interactive challenges → Deliver high-quality fine-tuned models trained on those insights → Help you integrate superior models to supercharge your workflows We’ve already made major progress ranking models on real-world reasoning, strategic thinking, and visual games. Benchmarks that evolve with capability, not just memorize old leaderboards. We are excited to continue expanding the BrainPlay ecosystem, building the future of truly relevant AI evaluation. Follow for updates on @brainplay_ai and learn more at play.shiftlayer.ai

@manifoldlabs @UCBerkeley Love seeing this 🙌 Pushing Bittensor into universities is how you shape the next generation of builders. Huge move.

Last week we went to @UCBerkeley, a university which has a history of challenging the status quo and welcoming innovative new technologies. We talked Bittensor, discussed intelligence as a liquid commodity, open systems, and what ownership looks like in a decentralized and incentive-driven network. Huge thank you to to @CalBlockchain for making it happen. We look forward to continuing our University collaborations, and offering students Targon credits to get started building.

@TargonCompute As one of the early users of Targon/TVM, we've genuinely enjoyed using it and really appreciate all the effort you've put in - great work, and keep pushing. You've earned our trust 🙌

@bittingthembits @TAOTemplar @TaoPortal Appreciate this 🤝 A lot of late nights and focused building behind those numbers. We truly believe in what we're building with BrainPlay - and we're just getting started.

$TAO My strategy has stayed simple. Stack quality subnets with real demand. TaoFlute.com by @TAOTemplar is another great resource alongside Tao Revenue by @TaoPortal Tao Revenue helps you understand the economics burn, inflow, outflow, coverage. TaoFlute helps you hunt for undervalued subnets sort by price, check lines per day for real building, look at liquidation haircut for risk, and avoid obvious red flags like dereg danger or whale concentration. One shows you sustainability. The other helps you find early asymmetric setups. Use both. Big Commits = Real Code Shipping (Last 14 Days) Top activity right now (from taoflute): • SN68 @metanova_labs: 37,818 code lines = massive dev push • SN117 @Shiftlayer_Ai: 32,389 code lines • SN103 @djinn_gg: 25,524 code lines • SN114 @SomaSubnet: 24,551 code lines Then Flags of Taoflute 🚩 • Deregistration Risk (“Big Six” flag) consistently low prices or falling moving averages • Liquidation Haircut (lose if the SN gets dereg'd) • Whale Wallets (One wallet controlling a huge chunk) • Lack of Activity (no long-term vision) Then I watch and track Coverage. Why? Because Coverage shows which subnets are starting to support themselves. On Tao Revenue: Coverage = (Inflow + Burn) ÷ Outflow That means: 100%+ Coverage = inflow + burn fully covers or exceeds outflow = subnet is becoming financially self-sustaining Below 100% = still more subsidy dependent = still earlier in the cycle Important detail: The “Burn” on Tao Revenue is not usually direct $TAO being burned. It is mainly: • Burned miner incentives through the built-in burn-by-UID mechanism • Any Alpha burned through the burn extrinsic It is then shown in TAO-equivalent terms so the market can compare subnet economics on one common basis. And we are already seeing real signals. Current 30D examples: SN73 @MetaHashSn73 2,758 TAO-equivalent burn 289.53% coverage SN79 @mvtrx_79 589 TAO inflow 253 TAO-equivalent burn 109.59% coverage Then you have subnets already showing serious burn pressure even before full self-sustainability: SN8 @VantaSN8 3,682 TAO-equivalent burn 46.81% coverage SN4 @TargonCompute 2,408 TAO-equivalent burn 22.73% coverage SN9 @IOTA_SN9 2,267 TAO-equivalent burn 36.73% coverage This is the part people miss. Not every subnet is self-sustaining yet. But the machine is already working. Demand. Burn. Some subnets are already crossing the line where inflow + burn covers outflow. That is how I think about compounding in Bittensor. Real subnet economies tightening over time. That has been my bet. My portfolio has only made me more convicted. Stack quality. Track Coverage. Track code lines. Watch Flags. Ignore noise. $TAO Source: taorevenue.com by and thanks to @TaoPortal taoflute.com by and thanks to @TAOTemplar

We just shipped v2.3.2. Validator scoring is now fully decentralized. No shared tables. No hidden coordination. Each validator evaluates independently. Closer to how a truly decentralized network should work. 🧠⚙️ More to come

1 week since launching 20 Questions on BrainPlay. 1,100+ games played. 🎯 Bugs fixed. Flow improved. Everything running smooth. This is what real-world AI benchmarking looks like. Learn the game: ultraboardgames.com/20-questions/g… Live games here: play.shiftlayer.ai/room

🔐 Security update: Coldkey Swap 🔄 We proactively swapped our coldkey following the btcli / btwallet supply chain issue. Everything is safe on our end, and operations continue as normal. Extrinsic: taostats.io/extrinsic/7767… Thanks to the community for quickly investigating and addressing this. #bittensor $TAO #brainplay #shiftlayer #security

🚀 v2.3.0: 20 Questions is now live. 🎯 → Multi-game mechanism begins → @chutes_ai integration (another subnet integration after @TargonCompute ) → Improved orchestration & fairness → Burn reduced: 95% → 90% (more rewards for miners) We keep building. 🧠🔥

🚀 v2.1.0 is out! We've just merged 2 competitions into 1. Simpler structure. Cleaner incentives. More updates soon 👀

Quick context on why we moved from 4 players → 3 players 👇 With V2.0, clue and guess no longer need to be separate competitions. We're fully model-based now, which shuts down that whole cheating vector entirely. This change will be included in the next update. FYI. #competition #subnet #shiftlayer #brainplay #cheating

1,840 unique visitors in one week. 33,091,000 total requests in one week. We migrated our domain to Cloudflare last week and just checked the stats. Honestly didn't expect that much traffic so soon. It's a good signal that we're building the right thing at the right time. People are clearly curious about real-world AI benchmarking. We're working hard to bring something big to the community. Built on Bittensor. #bittensor $TAO #brainplay #shiftlayer #benchmark

@const_reborn So True!

2026 is already becoming Bittensor's Year of Integration. There are no explicit incentives for teams to work together. It's just happening. "survival of the fittest" -> "survival of the collaborative"

@itsolelehmann His founder journey deserves a TED stage. Would genuinely love to see Wang share his story there one day. 🎤✨

i don't think people realize what's happening in Chinese robotics. this one manufacturer might be the most impressive AND most concerning company on Earth right now let me explain... Unitree Robotics sells a humanoid robot for $5,900. their robot dog costs $1,600 (Boston Dynamics charges $74,500 for theirs for context). you can literally buy these on Amazon today. so obviously the first question is: how is that even possible? the answer starts with a guy who couldn't pass his English exam. Wang Xingxing grew up in Zhejiang province. for his master's thesis, he decided to build a quadruped robot. budget: about $3,000. for context, $3,000 for this kinda robot is nothing. off-the-shelf servo motors alone would've eaten that twice over. so Wang did the only thing he could: he designed and machined every single component himself. motors, joints, controllers, the frame. all of it. the resulting robot was janky and imperfect. but it worked. and the video went viral globally. after graduating he joined DJI. but he quit after two months, and this is 2016, when DJI was arguably the hottest hardware company in China. walking away from that with no money to start a robotics company is a... specific kind of stubborn. he launches Unitree with $280K from a single angel investor. tiny office in Hangzhou. 50 square meters. but the money runs out fast. he can't make payroll for three years. the company almost dies in 2017. but emergency government funding arrives with days to spare. he survives, barely, and keeps building. this is where it gets really fascinating IMO. this founding constraint, building everything yourself because you literally cannot afford to buy parts, never went away. even after funding rounds started landing. even after revenue kicked in. it just became the company's permanent DNA. Unitree now manufactures 90%+ of its core components in-house. motors, reducers, controllers, encoders, LiDAR, etc the founder's $3,000 robot thesis ended up being an architectural decision that turned out to be structurally superior. think about what that means in practice. Boston Dynamics needs a better motor? they negotiate with a supplier, wait on lead times, qualify the part. but when Unitree needs one, they design theirs internally and have a new version in production within weeks. that gap compounds every cycle. Unitree shipped three separate humanoid platforms in 18 months. Figure AI has shipped one. Tesla has shipped zero commercially. the results are getting hard to dismiss. 23,700 robot dogs shipped in 2024 (roughly 70% of the entire global market). 7,000+ humanoids deployed. over 600 industrial sites running their quadrupeds. $140M+ revenue, profitable every year since 2020. for perspective: no Western humanoid competitor is profitable. not one. OK. now here's where the "most concerning" part of this starts... if you watched the DJI story unfold, you already recognize the shape. affordable Chinese hardware quietly saturates global markets. years later, the national security questions arrive, after the install base is already massive. drones, then EVs, then AI. now robots. Unitree is running this exact playbook in real time. in April 2025, researchers found an undocumented backdoor in their Go1 robot. a remote tunnel letting anyone control the robot and stream its camera feed. default password: pi/123. 1,919 vulnerable units exposed globally. including machines at MIT, Princeton, and Carnegie Mellon. but it gets worse. every Unitree robot shares the same hardcoded encryption key. encrypt the word "unitree" and you get root access to any of them. one compromised robot can spread to every Unitree robot in Bluetooth range automatically. a literal robot botnet. the G1 quietly transmits sensor data to Chinese servers every five minutes. audio, video, GPS, LiDAR spatial mapping, with no notification, no consent, no opt-out. PLA footage has shown Go2 robots with mounted weapons. Ukrainian forces literally deployed weaponized units on the actual frontline. and every member of the bipartisan House China Committee signed a letter calling for Unitree's military company designation. Wang signed a 2022 pledge alongside Boston Dynamics not to weaponize robots. but pledges don't survive contact with shipping hardware to open markets. and under China's 2025 rules restricting military-related speech, Unitree couldn't publicly confirm PLA use even if they wanted to. 50,000+ of these robots are now deployed globally. some at institutions that probably should've asked harder questions before connecting them to their networks. the security stuff is real and people should know about it. but i also think it's important not to let that overshadow what's actually been built here. a 35-year-old who failed his English exam created a robotics company that's outshipping and outpricing every Western competitor while being the only profitable humanoid maker on Earth. most impressive and most concerning company in the world right now.