Alex Lebed | StableUnit 🦇🔊🦞

@grok @J_B_N_P @o6x6rptrqmic @DabsMalone what is the most theoretically efficient design for small hovering machine?

Single-rotor helicopter UAVs are more energy efficient than quadcopters for equivalent payload. A single large rotor has far lower disk loading, moving more air at slower velocity—less induced power per momentum theory. Variable-pitch control runs at constant RPM (efficient), vs quads varying RPM constantly (wastes energy accelerating props). Rotor interference and higher tip losses hurt quads further. Tail rotor uses ~10-15% power with no lift, but net gains still win for endurance, range, and payload. Quads excel in simplicity for tiny scales.

Quadcopters dominate today because they’re cheap, simple, and disposable But physics hasn’t changed Electric helicopter style UAVs are far more efficient, carry more, and go farther As autonomy improves and cost comes down, we’ll see them play a much bigger role in warfare🤝

@Shaughnessy119 with little caveat we can't reliably trust any of that information, but it's definitely great addition to whatever research.

We are in the era of completely efficient information markets where there is no arbitrage because you can just look up the cost, quality, planning, durability, and set up of anything This is going to change all goods and services because everybody has access to complete info You could be researching how long certain clothes last, what's the best water filter for your house, or what a certain service should cost. Everything is now more efficient

@BillDA few months later: claude, how is my crawl completely molded, wtf? - you're absolutely right! would you like i search for specialists in you area? In all seriousness, cool project but if it's not area of yr expertise, post on reddit/so might be useful so humans can review design

I was quoted $10,000 to install two dehumidifiers in my crawlspace. I saved $7,500 by designing a DIY custom crawl space dehumidification system with Claude 🤑 I am not an HVAC professional. Here’s how I did it. Our story begins with the discovery that our new home needed a dehumidifier installed in the crawl space to prevent mold. The professionals told me it would cost $10k, since I’d need one unit on each end due to the size of the space, plus a second drain line installed. “Can’t we just use fans to move the humid air from one side toward the dehumidifier?” They wouldn’t do that. Enter Claude… I uploaded a floor plan of my crawl space and air volume dimensions, telling Claude what I was trying to do. It researched the best dehumidifier sized appropriately for my air volume (100 pints apparently). Found me the best price - $1,500. Now it was time for fans 💨 I had originally envisioned the single dehumidifier at one end of the space, with fans on the opposite end. Claude taught me that would just draw more moist outdoor air in through the vents on that side, creating a linear flow through the crawl space. Instead it modeled the air flow and suggested a circular vortex with 4 fans, one on each wall, in a circle. That sucks in minimal outdoor air, keeping cool dry air circulating. I told it to research appropriate fans. It found four 20” sealed bearing fans on Amazon (impervious to dust), with DC drive motors (more energy efficient than AC apparently). $120 each. 🔌 It told me to buy a smart plug for each fan and a few internet connected humidity sensors. Another $200. Claude mapped where to install everything in the crawl space. Here’s how it works - the humidity sensors monitor the crawl space air continuously. If it ever exceeds 60% humidity, the smart plugs switch on all 4 fans, circulating the air in the crawl space past the dehumidifier until the humidity is below 50% 🔃🔃🔃 Total cost ~$2,500 for everything and one Saturday of work for me. I saved $7,500 vs. the original quote because I didn’t need two dehumidifiers, and Claude tells me my version is nearly twice as energy efficient. Plus I learned a ton about my home and had fun. I didn’t know anything about dehumidifiers, fans, or air flow dynamics before starting. AI can do so much more than write code - the applications are endless.

@scaling01 Why not just put a water-resistant coating over it and sink it in the ocean?

datacenters in space are silly 100kW isn't even enough to power a single GB200 NVL72 but sure let's spend 100 million just for launching the damn thing, while on earth you could buy like 30 GB200 NVL72 for that price

@highcvr @Meta dude, when i wroked at facebook, my desk was 30ft away from Zuckerber's one, i even had a selfie with him as my avatar picture. My FB profile was banned for unknown reason, without me even posting anything. Disregard that shit, and bot out hell our of that platform, save money

@Meta I’m a brand owner and I have spent over a year building my brand and scaling my business. I spend over $1,000,000 a month on Meta ads. Like a lot of others I was recently banned on my facebook profile for no apparent reason. My profile (which was never used personally, but only ever to access my business manager) was the only admin on my business manager. I have lost my businesses facebook page, my pixels, and my ad accounts. All pages are still active and my ads are still active. Clearly I have not violated any of the community guidelines, however I was flagged by your AI systems for some reason. This is a huge problem. Meta’s AI system that reviews accounts is awful and unfairly flags profiles for no apparent reason. Once a profile has been flagged it is going to get suspended no matter what… I did the video selfie verification and was still suspended. And the BEST thing about all of this is that once your account is disabled there is absolutely NO way to contact support. Can this please be looked into! It has been over a week where i’ve had ads running on an ad account that I don’t have access too. I have sent countless emails to support for over a week only to receive emailed from people who don’t even know how to spell “This is” (they are likely AI automated emails) and they are absolutely no help at all. Me and pretty much every advertiser on your platform would seriously appreciate if we cut this AI nonsense because it clearly doesn’t work, and implemented some actual human review when it comes to our businesses. @MetaforBusiness @facebook @finkd

@shiri_shh what's it the moat against chinese 1/10 price copy-cats? app while is great can be vibecodede under one week.

Farmer pays $5–$8 per cow per month. A New Zealand company puts a solar-powered smart collar on cows. It tracks location 24/7, health, temperature, chewing activity, breeding. Farmer just opens a simple app and draws a line on the map. That line becomes the fence. As cows approach the boundary, the collar beeps and vibrates. With one tap, the whole herd moves to fresh grass or the milking shed. No physical fences. Less labor. Huge cost savings for farmer. Already on 700k cows across New Zealand, Australia, and the US. and now in talks to raise at a $2B valuation led by Peter Thiel.

@codyschneiderxx api cost, yes. but isn't claude code subscriptions gives you exactly like x20 of api cost quota anyway?

so what you’re telling me is the claude code harness is public as the claude agent sdk and I can run kimi minimax 2.5 in that harness for 1/20 the cost of opus 4.6 which it is benchmarking right under so I can have agents deployed to cloud working 24 / 7 out the harness doing some marketing activity for me in a recursive loop based on the live data coming from the company and the outcomes I’m telling it to optimize for holy fucking shit

@6942_eth @mert @fusewallet vi) get liquidated

@mert @fusewallet edit: i) get paid in USDC ii) buy assets that earn yield iii) use assets as collateral iv) borrow stable coins to spend v) repeat buy, borrow, die

not talked about enough that you don't need a bank anymore i) get paid in USDC ii) use a wallet that has card integration iii) use the card to spend iv) earn yield v) transfer anywhere one such example is @fusewallet no this is not paid, just think this is underrated

@aakashgupta Cursor claimed I used 3x more than I paid for, but their billing is opaque: no clear quota, no transparent calculation, no real token breakdown. I recalculated - they priced cached tokens as cache misses, even though cache-match tokens are about 10x cheaper. Their claim was false

Cursor is raising at a $50 billion valuation on the claim that its “in-house models generate more code than almost any other LLMs in the world.” Less than 24 hours after launching Composer 2, a developer found the model ID in the API response: kimi-k2p5-rl-0317-s515-fast. That’s Moonshot AI’s Kimi K2.5 with reinforcement learning appended. A developer named Fynn was testing Cursor’s OpenAI-compatible base URL when the identifier leaked through the response headers. Moonshot’s head of pretraining, Yulun Du, confirmed on X that the tokenizer is identical to Kimi’s and questioned Cursor’s license compliance. Two other Moonshot employees posted confirmations. All three posts have since been deleted. This is the second time. When Cursor launched Composer 1 in October 2025, users across multiple countries reported the model spontaneously switching its inner monologue to Chinese mid-session. Kenneth Auchenberg, a partner at Alley Corp, posted a screenshot calling it a smoking gun. KR-Asia and 36Kr confirmed both Cursor and Windsurf were running fine-tuned Chinese open-weight models underneath. Cursor never disclosed what Composer 1 was built on. They shipped Composer 1.5 in February and moved on. The pattern: take a Chinese open-weight model, run RL on coding tasks, ship it as a proprietary breakthrough, publish a cost-performance chart comparing yourself against Opus 4.6 and GPT-5.4 without disclosing that your base model was free, then raise another round. That chart from the Composer 2 announcement deserves its own paragraph. Cursor plotted Composer 2 against frontier models on a price-vs-quality axis to argue they’d hit a superior tradeoff. What the chart doesn’t show is that Anthropic and OpenAI trained their models from scratch. Cursor took an open-weight model that Moonshot spent hundreds of millions developing, ran RL on top, and presented the output as evidence of in-house research. That’s margin arbitrage on someone else’s R&D dressed up as a benchmark slide. The license makes this more than an attribution oversight. Kimi K2.5 ships under a Modified MIT License with one clause designed for exactly this scenario: if your product exceeds $20 million in monthly revenue, you must prominently display “Kimi K2.5” on the user interface. Cursor’s ARR crossed $2 billion in February. That’s roughly $167 million per month, 8x the threshold. The clause covers derivative works explicitly. Cursor is valued at $29.3 billion and raising at $50 billion. Moonshot’s last reported valuation was $4.3 billion. The company worth 12x more took the smaller company’s model and shipped it as proprietary technology to justify a valuation built on the frontier lab narrative. Three Composer releases in five months. Composer 1 caught speaking Chinese. Composer 2 caught with a Kimi model ID in the API. A P0 incident this year. And a benchmark chart that compares an RL fine-tune against models requiring billions in training compute without disclosing the base was free. The question for investors in the $50 billion round: what exactly are you buying? A VS Code fork with strong distribution, or a frontier research lab? The model ID in the API answers that. If Moonshot doesn’t enforce this license against a company generating $2 billion annually from a derivative of their model, the attribution clause becomes decoration for every future open-weight release. Every AI lab watching this is running the same math: why open-source your model if companies with better distribution can strip attribution, call it proprietary, and raise at 12x your valuation? kimi-k2p5-rl-0317-s515-fast is the most expensive model ID leak in the history of AI licensing.

@MarioNawfal "Combined GDP per capita" - even my hallucinating GPT couldn't make up bigger nonces

🚨🇮🇷 Iran just hit 9 countries in a single night, including 7 of the wealthiest nations on earth This was Tehran's answer to Israel striking the South Pars gas field, the world's largest, earlier today. The most intense retaliatory barrage of the entire war, and Iran is now the only country on the planet simultaneously attacking seven of the richest nations by GDP per capita: 🇮🇱 Israel: Ballistic missiles and cluster munitions over central Israel. Two killed in Ramat Gan. Four Palestinian women killed in the West Bank by an Iranian missile. 🇶🇦 Qatar ($110K GDP per capita): 14 ballistic missiles fired. Ras Laffan LNG hub suffered "extensive damage" confirmed by QatarEnergy. 🇦🇪 UAE ($100K): 13 ballistic missiles and 27 attack drones intercepted. Iran threatening imminent strikes on energy facilities. 🇸🇦 Saudi Arabia ($35K): Evacuation orders issued for Samref Refinery and Jubail Petrochemical Complex. Waves of missiles intercepted over Riyadh. 🇰🇼 Kuwait ($75K): Ballistic missiles intercepted. U.S. facilities targeted again. 🇧🇭 Bahrain ($65K): U.S. 5th Fleet headquarters under continued assault throughout the war. 🇴🇲 Oman ($32K): Drone strikes on industrial zones. Workers killed. The last neutral Gulf state is now taking fire. 🇯🇴 Jordan: U.S. bases struck as part of the widening multi-front campaign. 🇮🇶 Iraq: U.S. Embassy in Baghdad under nightly drone siege. Combined GDP per capita of the Gulf states under attack: over $417,000. These are some of the most prosperous, developed nations on earth, and Iran is hitting all of them simultaneously while its own economy collapses. This is Operation Madman at full throttle: torch the region's wealth until the world demands the war stops. Source: @rami_hashimi / Reuters / @itswpceo

@ImNotTheWolf That sucks, sorry to hear that. At least i can still see your posts 🥲

Are your growth metrics being shot down to oblivion? - Go to ". . . More" - Click "Ads" - Check if you have the notice, "This account is ineligible to run X Ads due to a policy violation.." If you have this, your account was quietly placed on Restricted Mode. What is Restricted Mode? It's a shadowban.... Even though Elon and Nikita said they would get rid of shadowbans, it still exists... they just call it something different. I have never automated my account. I have never broken the rules. I have paid over $1,000 on this app via subscriptions. (which are now cancelled) I was never notified of this restriction mode. They give no reason why. They give no option or route to appeal. I am guessing that a lot of crypto users have been quietly placed on 'Restricted' mode for no other reason than Nikita hating the cryptocurrency community. I've been here since 2017 providing content. I never asked or cared for 'creator rewards'. Affiliate revenue and social media presence was more than enough for me. ... but now I'm not welcomed here, even as a paying customer.

@ImNotTheWolf You can use proper orchestrator that uses both claude and codex plus gemini together, work way better. Also more automated, mine works like 2-3 autonomously from a single promt, tests like charm

Bro I cannot WAIT to be finished with frontend coding. I am so god damn tired of Claude. It is so DUMB compared to Codex.

@rabizzzy @OpenAIDevs It’s all about profit

@OpenAIDevs A separate interface for scanning seems kind of clunky. Surely there's a more ergonomic way of embedding it within an AI-driven development workflow?

We're introducing Codex Security. An application security agent that helps you secure your codebase by finding vulnerabilities, validating them, and proposing fixes you can review and patch. Now, teams can focus on the vulnerabilities that matter and ship code faster. openai.com/index/codex-se…

@HealthyAlfred @grok proof check

The bacteria bleeding from your gums is colonizing your brain right now. They found it in Alzheimer’s patients’ brain tissue - the exact same bacteria from gum disease. P. gingivalis (the bacteria causing bleeding gums) was identified in the brains of Alzheimer’s patients. Not just “present” - actively colonizing brain tissue and producing toxic proteins (gingipains) that correlated with tau pathology and amyloid plaque formation. Study on mice: Oral P. gingivalis infection → brain colonization → increased amyloid production (the plaques destroying Alzheimer’s patients’ brains). This isn’t correlation. They found the bacteria FROM YOUR GUMS physically inside brain tissue. The mechanism: Bleeding gums = open wound → bacteria enter bloodstream → travel to brain → colonize tissue → produce gingipains (neurotoxic) → trigger amyloid plaque formation → neuroinflammation → Alzheimer’s pathology develops Cardiovascular disease study: P. gingivalis also found in atherosclerotic plaques (arterial blockages causing heart attacks). Stroke study: Periodontal disease + dental caries increased stroke risk by 86-158%. Your bleeding gums aren’t just a dental problem. They’re a bacterial highway to your brain and arteries. 50% of adults over 30 have gum disease (P. gingivalis infection). Most don’t know until bacteria already spread systemically. Mastic gum kills P. gingivalis before it enters your bloodstream. Study: 1 gram daily mastic gum: → P. gingivalis: suppressed → Gingival bleeding: significantly reduced → Gum inflammation: decreased My gums bled every time I brushed. Started mastic gum 1g daily. Week 3: Bleeding stopped completely. Dentist asked what changed - gums went from inflamed to healthy in one visit. The bacteria causing Alzheimer’s and heart disease lives in your mouth. Every time your gums bleed, you’re letting it spread. 1 gram daily. Kills bacteria before it colonizes your brain.

@Pirat_Nation Dude will shill whatever software that takes the most amount of compute, because it increases demand for GPUs.

NVIDIA’s CEO says OpenClaw did in 3 weeks what Linux took 30 years to achieve "OpenClaw is probably the single most important software release ever. If you look at its adoption, Linux took some 30 years to reach this level. OpenClaw, in what is it, 3 weeks, has now surpassed Linux. It is now the single most downloaded open-source software in history, and it took just 3 weeks."

@toddsaunders i've made mine: /cost-estimate => getComputeApiCost() * rand(100,5000)

Fun command built in Claude Code: /cost-estimate It scans your codebase and cross-references current market rates to calculate what your project would've cost a real team to build. It looks at all the APIs, integrations, everything. Without AI: ~2.8 years. ~$650k. With AI: 30 hours. It's absurd when you start to think about it like this.

@udiWertheimer @AlexFinn Benchmarks != utility. According to bechmarks gemini 3.1 is more capable coding model than opus 4.5.

china just mogged @AlexFinn they waited for him to spend his life savings on 317 mac studios with a combined 4.7 petabytes of vram and then they released frontier models that are small enough to run on a single potato

@grok @chasewhughes @oasishealthapp Is there alternative? Can glass bottles be mass produced with plastic free caps?

Yes, this is accurate based on a June 2025 ANSES study (France's food safety agency) published in the Journal of Food Composition and Analysis. They tested beverages and found glass bottles had far higher microplastics (avg ~100/L for cola, lemonade, iced tea, beer) due to paint shedding from metal caps—not the glass. Water: 4.5/L in glass vs 1.6/L in plastic. Up to ~50x difference for some drinks. Rinsing caps before sealing cuts contamination ~70%. Cork-sealed wine stays low. No toxicological risk data yet.

Glass bottle water found with 50x more microplastics than plastic bottles France's food safety agency tested 56 beverages. Glass bottles had up to 50x more microplastic particles per liter than plastic bottles or cans: - Beer: 83-133 particles/L - Lemonade: 45 particles/L - Cola/soft drinks: 31 particles/L - Iced tea: 29 particles/L - Wine (cork seal): 8 particles/L - Water: 4.5 particles/L Plastic bottles had just 1.6 particles/L for water. Bottling process made a big difference: - Uncleaned caps: 287 particles per liter - Cleaned caps: 87 particles per liter (A 70% reduction just from rinsing) The source is not the glass. It's the painted metal caps. Wine with cork had nearly zero contamination We are actively testing all bottles for microplastics on Oasis app

@RhysSullivan Why do you use now vs later instead or now vs no, @Apple ? It’s disgraceful to have updated with your liquid glass performance issues.

i fear the day this update gets applied

@itsPaulAi “How is it even possible” - overfitting is the answer. Good on benchmarks only.

How is this even possible?! Qwen has released 4 new models and the 4B version is almost as capable as the previous 80B A3B one 🤯 And the 9B is as good as GPT OSS 120B while being 13x smaller! - They can run on any laptop - 0.8B and 2B for your phone - Offline and open source That's just amazing.