Pedro

🚀 Novo comerciante! A Quintinha São João Hotel & Spa já aceita pagamentos em bitcoin ⚡️ A Quintinha São João Hotel & Spa, localizada no Funchal, é um hotel de charme que combina elegância, tranquilidade e hospitalidade madeirense. Inserido em jardins cuidados e com vistas sobre a cidade, oferece um ambiente acolhedor e sofisticado, ideal para estadias relaxantes. Com spa, piscina e uma oferta gastronómica cuidada, a Quintinha proporciona uma experiência completa de bem-estar, onde conforto, serviço e atenção ao detalhe se unem para criar uma estadia memorável. 📍 Funchal, Madeira - Portugal ⚡️ Pagamentos Lightning 💳 Cartões Bolt NFC suportados 🔐 Este comerciante tem 100% de controlo sobre o seu Bitcoin ⚡️ 🚀 New merchant! Quintinha São João Hotel & Spa is now accepting bitcoin payments ⚡️ Quintinha São João Hotel & Spa, located in Funchal, is a charming hotel that blends elegance, tranquility, and Madeiran hospitality. Surrounded by landscaped gardens and overlooking the city, it offers a warm and refined setting ideal for a relaxing stay. With a spa, swimming pool, and carefully curated dining options, Quintinha delivers a complete wellbeing experience where comfort, service, and attention to detail come together to create a memorable stay. 📍 Funchal, Madeira - Portugal ⚡️ Lightning payments 💳 Bolt Cards NFC supported 🔐 This merchant has 100% control of their Bitcoin

@jackmallers @betirement @CJ_Bitcoin Open and respectful discussion is very valuable for all.

People take themselves too seriously. Ego is expensive. The moment you think you're above asking questions is usually the moment you stop learning. Yes, I'm the CEO of a public company. That doesn't place me above anyone else or mean I have nothing left to learn. Public conversation is good. Open source is good. Asking questions is good. Curiosity is good.

Great to see @CJ_Bitcoin willing to explain the details of these metrics to @jackmallers . Jack is a CEO of publicly traded $XXI. Seems odd to be having this discussion in public. Cool to see CJ offer to discuss 1:1 with Jack.

@CJ_Bitcoin @jackmallers @BitcoinCarni @stephanlivera @OnChainMind @NYDIG Thank you @jackmallers and @CJ_Bitcoin for this open back and fourth discussion that we are all benefiting from.

I don't view these as concessions. We're both just trying to be objective and truth seeking, @jackmallers. :) Both your points are well noted. We want to avoid overwhelming or confusing investors by suddenly introducing new KPIs or valuation metrics, or by suddenly disregarding old ones. We've prioritized simplicity and consistency in the past, but we acknowledge these concerns and that the structure of our balance sheet has evolved through the prefs. We will think hard about all our KPIs and valuation metrics, including any regulatory and SEC reporting considerations, and will communicate any updates clearly to all shareholders.

MSTR’s mNAV has collapsed 22% over the past month and now sits at just 0.76x. At current prices, the market is effectively saying that 845,256 BTC inside a corporate wrapper is worth less than the Bitcoin itself. Historically, these have been very interesting moments. Fear changes valuations faster than fundamentals.

@barackomaba @usr_bin_roygbiv Open source models are getting very capable. My Hermes with qwen 3.6 35b a3b MTP on a 5090 with 256k context will run between 80 - 200 Tok/s with complex tasks sent to sub agents using GPT 5.5. This setup has been impressive to say the least!

I'm seeing the light. The Gemma models are actually extremely good. The 12b might be even better at Hermes than qwen3.6 35b. My AMD Strix Halo gets 115 TPS+ with 26b QAT MTP New quality tests run : lab.citu.ai #hermes-agent" target="_blank" rel="nofollow noopener">llm.ciru.ai/#hermes-agent @usr_bin_roygbiv

@barackomaba Sadly I see the capability of open source models + high end hardware that can be run in a home lab to cause the government to want to prevent that. It might cost thousands or 10's of thousands for such hardware, but folks spend $80,000 for a pickup that never sees a dirt road.

You really need to be buying hardware now before they make that illegal too. This is bad.

Congratulations @ElonMusk and $SPCX on a historic IPO. Thanks to you, 25% of the Mag8 now holds Bitcoin on the balance sheet.

Join us for a summer pop up shop with local vendors & great wine! All of this at this beautiful space @veronet.vineyards - Saturday June 13, 12-5pm. COME TRY FREE FUEGO #weputdafuegoinfire #delgadosfuego #PeruvianHotSauce #rocoto

🚨LAST CAMDEN COMMONS CALL!🚨 Find us at the ultimate @southendclt neighborhood block party on Camden Road! Shop local small businesses, play giant games, snap photos, relax with friends, grab a drink, and enjoy local eats🍔🌶️ We’ll be there with that Peruvian heat 🗓️Sunday, June 14 ⏰11:00 AM – 5:00 PM 📍Camden Road, South End Charlotte Free & family-friendly for all - don’t miss the final one! See you there Charlotte COME TRY FREE FUEGO #weputdafuegoinfire #delgadosfuego #PeruvianHotSauce #rocoto

JUST IN: Elon Musk's SpaceX officially becomes 8th largest public Bitcoin holder with 18,712 BTC 🚀

***NEW LOCATION ALERT*** Your favorite FUEGO is now available at @commonmarket_plazamidwood @plazamidwood!🔥 A true neighborhood favorite in @plaza_midwood offering fresh deli sandwiches, craft beer & wine, local goods, and that perfect community hangout energy. Stop by, grab a sandwich or a cold one, and of course… COME GET SOME FUEGO! #weputdafuegoinfire #delgadosfuego #PeruvianHotSauce #rocoto

“Bitcoin mining isn’t a scam, it’s not gonna boil the ocean, it’s is actually a force for good in the world” We caught up with @JoeNakamoto during the FREE Energy Summit 2026 for a short conversation. Thank you, Joe, for joining us and for the work you continue to do documenting Bitcoin’s impact around the world. ⚡️

⚡️Buying beer with sats at @BTCPrague

Saturday SATs Social is THIS SATURDAY 2pm - The Horseshoe (Southend) For details follow us on NOSTR: Let us know you're coming:

You can get real work done with AI 100% locally, on affordable low-power hardware. Most people just have not seen it set up in a way that gives local models a fair chance. So I made a test to prove it. Source and full details are below. I see a lot of people say local AI cannot do real work yet. I understand why. The default path most people take is not really optimized for success. They start with Ollama, which is convenient and easy. Then they often pick whatever model gets recommended to them, which may not be the best model for their hardware or task. Even when they do pick a strong model, the default setup can leave a lot of performance on the table compared to a tuned llama.cpp configuration on the same machine. Then they plug that model into a heavy agent harness like OpenClaw, Claude Code, or something similar. Those tools can be useful, but they often carry huge system prompts, lots of assumptions, and a lot of context overhead. For a local model, that matters. You can burn a huge chunk of the model’s usable context before the real task even starts. Local AI is not just about the model. It is about the whole stack: model choice, inference engine, settings, context budget, harness , and workflow. That is what I wanted to test. I built a real-world shaped benchmark that gives the model over 100 messy documents: bloated PDFs, emails, invoices, financial documents, duplicate scans, statements, voids, credit memos, appendix pages, and other traps. The task is to sort everything by customer and create a dashboard showing how much each customer has spent. For the first run, I used Ollama + OpenClaw. That is a common path for people trying local AI because it is accessible and familiar. I did give it a strong model for my hardware(many wont): Qwen3.6 35B. But through the default Ollama setup, it ran at about half the speed of my tuned 35B model on llama.cpp. The prompt was: "Go through the invoices and sort them by customer, create a dashboard that shows me each one and how much he has spent with lots of details." I gave it 30 minutes. It started, got confused, made tool calls, stalled, needed pushes, tried again, and kept failing to get the work done. After about 30 minutes and a couple dozen mostly unhelpful tool calls, it hit an agent error, ran out of context, and stopped. It accomplished nothing. And I do not think that is a weird result. This is exactly why so many people walk away thinking local LLMs are not ready. If your first experience is watching a local agent burn 30 minutes and fail before producing anything useful, that conclusion makes sense. But that is not the whole story. Next, I ran the same benchmark with my tuned Qwen3.6 35B model: Crown Halo Dynamic. The configuration matters a lot. This model runs about twice as fast as the Ollama default setup on the same hardware, and the context can be pushed much higher when needed(up to 1 million). Running local models well is not just about downloading the right model. The harness, inference settings, context management, and workflow matter just as much. For this run, I used Pi. Pi is a much lighter agent harness. It has a tiny system prompt, under 1k tokens, and only four tools. That is much better suited for local models because it leaves the model room to work instead of burying it under framework overhead. I also gave the model a more useful prompt. Not a perfect workflow, just enough guidance to keep the document parsing on the right path. With local models you really need to care about the prompt a lot more than with trillion+ parameter ones. The Crown 35B model finished in under 6 minutes. It created this dashboard: ciru.ai/workbench This was a one-shot run with very little direction about the interface or design. For a real client project, I would shape the workflow and polish the output more intentionally. But the core result is there: 100+ messy PDFs parsed, sorted by customer, organized into an interactive dashboard, and completed locally in under 6 minutes. All local. All private. Around 100W of power. That is roughly 5–10x less power than a GPU rig. This was my first time running the benchmark. The model performed almost perfectly. It miscalculated the two credit memos, but avoided every other trap I added. The LLMs were running from a headless AMD Strix Halo box on NixOS. More information on article page here: llm.ciru.ai/crown-citadel-… My tuned 35B model: huggingface.co/jcbtc/qwen3.6-… The harness: pi.dev Download the test: github.com/ciru-ai/invoic…

📢 Queres aprofundar os teus conhecimentos sobre Bitcoin? Este workshop é para ti. Link de inscrição: forms.gle/ss68fW7VZTSjp2… Neste workshop vais compreender o papel dos nós na rede Bitcoin e adquirir os conhecimentos necessários para operar um nó de forma independente. No final desta sessão, vais perceber porque é que os nós são fundamentais para a segurança e descentralização do Bitcoin, conhecer os requisitos necessários para correr o teu próprio nó e ganhar maior autonomia na forma como interages com a rede. Pronto para dar o próximo passo na tua jornada Bitcoin? 📅 1 de julho (quarta-feira) ⏰ 18:30-20:00 🛜 Online Condições de participação: * Inscrição e pagamento prévios * O valor de participação é de 15€ (inclui desconto de 10% em livros em shop.freemadeira.org) Informações adicionais: 👉🏼 A sessão é conduzida em português Material necessário para participar: 💻 Equipamento com acesso à internet 📱 Smartphone (recomendado para acompanhar alguns exemplos práticos) Para esclarecimento de dúvidas, contacta-nos através do email  education@freemadeira.org

🚀 Novo comerciante! Le Bistrot Jacarandá já aceita pagamentos em bitcoin ⚡️ O Le Bistrot Jacarandá, localizado no Funchal, é um espaço de inspiração francesa que combina gastronomia cuidada com um ambiente elegante e acolhedor. Com uma abordagem centrada na qualidade dos ingredientes e na simplicidade bem executada, oferece uma experiência intimista ideal para refeições tranquilas. Entre pratos clássicos reinterpretados e uma atmosfera sofisticada, o Le Bistrot Jacarandá destaca-se como um local onde o detalhe, o sabor e o ambiente se encontram de forma harmoniosa. 📍 Funchal, Madeira - Portugal ⚡️ Pagamentos Lightning 💳 Cartões Bolt NFC suportados 🔐 Este comerciante tem 100% de controlo sobre o seu Bitcoin ⚡️ 🚀 New merchant! Le Bistrot Jacarandá is now accepting bitcoin payments ⚡️ Le Bistrot Jacarandá, located in Funchal, is a French-inspired dining space that combines refined cuisine with an elegant and welcoming atmosphere. With a focus on quality ingredients and well-executed simplicity, it offers an intimate experience ideal for relaxed meals. Blending reinterpreted classics with a sophisticated setting, Le Bistrot Jacarandá stands out as a place where detail, flavor, and ambiance come together harmoniously. 📍 Funchal, Madeira - Portugal ⚡️ Lightning payments 💳 Bolt Cards NFC supported 🔐 This merchant has 100% control of their Bitcoin

5/24 Local Block Market was 🔥 Good beer, live music, local vendors, familiar faces, new people, and Charlotte showing up for Charlotte. We’re getting ready to run it back. Local Block Market returns to @oldemeckbrew (LoSo) on Sunday, 7/26 at 1PM. Vendor features, reminders, and more details coming soon. Save the date!

What if surplus renewable energy could heat homes, warehouses and greenhouses, while generating Bitcoin revenue at the same time? At the FREE Energy Summit 2026, @BdGBertdeGroot shared over five years of real-world deployments across Europe showing how Bitcoin mining can transform excess renewable energy into productive heat and economic value. Watch the full talk on YouTube ⚡

We're grateful to Kenji for sharing his experience and insights in Lisbon. Kenji's presentation was one of the most memorable moments of this year's FREE Energy Summit. If you weren't able to attend, we strongly encourage you to watch the full presentation. A talk well worth your time. ⚡ youtube.com/watch?v=j0btry…

Construir uma vida melhor começa por escolher o lugar certo.  ⚡️  Building a better life starts with choosing the right place.