WOW! 🤯 Language models are becoming smaller and more... capable than ever! Here's SmolLM2 running 100% locally in-browser w/ WebGPU on a 6-year-old GPU. Look at that speed! ⚡️😍 Powered by 🤗 Transformers.js and ONNX Runtime Web! How many tokens/second do you get? Let me know! 👇show more

HOLY MOLY running a 35B model locally on a... MacBook shouldn’t be THIS FAST 🤯 Spent my weekend in atomic.chat testing Qwen 35B vs. Qwen 27B on my local machine. I had them generate a fully animated HTML/Canvas car mini-game (demo below), ... and both models breezed through the physics and parallax scrolling without a hitch! The secret sauce here is the Atomic Chat app. Because it's perfectly optimized for Mac and uses Google's new TurboQuant under the hood, you can run heavy open-source models flawlessly while keeping top-tier output quality 👊 Other perks: → ZERO setup required → Access 1,000+ models completely free → 100% offline and private → Zero API limits ... and MUCH more! I dropped the prompt I used in the 🧵↓ Spin it up locally and let me know what you get!show more

Sub-40ms full self-driving on a $100 drone! 🚁 A... demonstration showing a complete full-self-driving pipeline running in under 40 milliseconds on a $100 drone. The prompt: "Find the bike and land." No pre-mapping. Running real-time on commodity hardware. For context, human reaction time is around 200-250ms. This drone is processing sensor data, understanding natural language commands, identifying objects, planning motion, and executing control, all in 40ms. This is what happens when foundation models meet efficient inference. The models get smaller and faster while maintaining capability. The hardware gets cheaper while getting more powerful. The intersection makes previously impossible applications suddenly viable. A few years ago, this required thousands of dollars in compute, pre-mapped environments, and cloud connectivity. Now it runs locally on hardware that costs less than a nice dinner. Awesome stuff Chester & ! 😮‍💨 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →show more

you're paying $20/mo for something your $500 GPU can... already do. Gemma 4 26B A4B QAT MoE + Hermes Agent running on a single RTX 4060 (8GB VRAM). Built a vision capable, 100% free, 100% local, private AI assistant that lives in my Chrome browser. No API keys. No cloud. No subscriptions. 100% vibe coded. 0% handholding. It has full context of whatever's on my screen can answer questions, summarize pages, extract data, and see images. Same local model handles everything, no external calls, ever. keep reading for the model and hermes agent tips i learnt while building this locally. Here's the exact setup for anyone running local LLMs on 6-8 GB VRAM: llama.cpp server flags (on my NVIDIA RTX 4060 8gb VRAM): -m gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf --cache-type-k q8_0 --cache-type-v q8_0 -c 150000 --port 8080 Throughput with quantization: Prefill: 200-250 tokens/sec Decode: 20-25 tokens/sec reduce context if oom on 6 gb vram card. Key learnings: - Quantize KV cache to q8 for faster prefill/decode. Prefill goes from 100-150 (unquantized) to 200-250 tok/s (q8). - But watch out, once actual context grows past ~50k tokens on high entropy workloads, q8 KV quantization can cause hallucinations. Low entropy workloads are mostly unaffected. If you see it happening, drop the quantization. This is common across all local models. - In Hermes Agent settings -> Memory & Context, bump compression threshold from default 0.5 to 0.7. Default triggers way too frequent context compression and eats time. Up next: add persistent memory, web search, tool calling, streaming output and whatever you suggest. Running a 26B MoE with vision + 150k context window on 8GB VRAM would've sounded impossible 6 months ago. Works the same on the NVIDIA RTX 3060 Ti, 3070, 4060 Ti, 5060, 2080, or any 8GB card. VRAM is the only requirement. Local AI agents are closer than people think. You just need to know where the knobs are. Model's Unsloth quant hugging face link in the comments. Have you tried Hermes agent by Nous Research yet? What are you building with local LLMs? Drop it below, let's see what this community is shipping.show more

Testing the new Gemma 4 12B (QAT) vision and... OCR capabilities locally with LM Studio. # The setup: - GPU: NVIDIA RTX 4060 (8GB VRAM) - CPU: Intel i7 - Runner: LM Studio - Config: 32k context, 38 layers offloaded, Flash Attention enabled - Speed: ~14 tokens/sec decode throughput # The test: I gave it a screenshot of Google AI Studio. Prompt: "clone this. give me a single html file" # The result: A solid one shot replication. It successfully mapped out the layout, recognized the UI text, and structured the divs correctly, with only minor differences from the original. Results available at the end of the video. Quite capable for a 12B model running on budget consumer hardware. A gpu that costs only $300. # Why the architecture under the hood is notable: Unlike traditional models that rely on heavy, separate vision and audio encoders, Gemma 4 12B uses a unified, encoder free architecture. It bypasses separate multi stage encoders. Uses a 35M parameter vision embedder to project raw 48x48 pixel patches directly to the LLM hidden dimension. Local multimodal development is becoming highly accessible on standard hardware. If you've spun up Gemma 4 12B locally, what setup are you using and what kind of throughput are you seeing?show more

"He told them he'd be back by dinner. That... was 14 months ago." Every single evening, Max and Buddy walk to that back door at exactly 5:47pm. They don't know what a deployment is. They don't understand overseas. They just know his smell is fading from his old jacket on the hook. And they keep waiting anyway. My brother asked me to look after them "just for a few months." I didn't know I'd fall in love with two dogs who taught me more about loyalty than any person ever has. If you've ever loved an animal who waited for someone — you know this feeling in your chest right now. Drop a ❤️ if Max and Buddy deserve to see their dad come home soon.show more

5,000 of you. I don’t even know how to... put that into words. When I started sharing my art and stories here on X, I honestly didn’t know if anyone would care. The AI community can be loud, fast, overwhelming… and yet, you made space for me. You saw my work. You shared it. You quoted it. You commented. You encouraged me when I doubted myself. This feels like more than a number. It feels like a second chance. I truly believe this might be my last real shot at showing my art and stories to the world, and the fact that 5,000 of you chose to follow, support, and believe in what I create means more than I can explain. Every repost, every like, every thoughtful reply has carried me further than you know. Thank you for standing by an artist experimenting at the edge of AI and imagination. Thank you for helping my stories breathe. Thank you for being here. We’re just getting started.show more

🔥IVY TOKENS WILL BE BURNED TO MAKE IT THE... FIRST DEFLATIONARY MEME ON HEDERA🔥 IMPORTANT: The $IVY token will be deflationary Yep, IVY, the guardian of the hashgraph, just announced one of its first token burn sources, in partnership with ĦONKbot Read the announcement carefully, and in the next tweets, we'll explain how to use HonkBot First off, we wanna say that we're designing several burn sources for IVY tokens. Here's the first one 🌱🔥 On all chains, Telegram bots are pretty popular, from Banana on Ethereum to Trojan on Solana. They're handy tools that let you buy assets easily, right from your Telegram. By signing up with the code for HonkBot, ALL the fees generated from referrals will be used to buy back IVY tokens and send that amount to a burn wallet 💰 So, the more you use the bot, the more fees are generated, which means more IVY buybacks and more tokens sent to be burned This is just one of the many burn sources we've got planned for IVY. If you don't know how to use a Telegram bot, don't sweat it, we'll walk you through it next. Watch the video and turn up the volume.show more

Run Gemma 4 26B MoE on 8GB VRAM with... 250k context at 20+ tokens/sec If you own any 8GB VRAM graphics card, stop what you are doing. Local AI just had its absolute "Holy Shit" moment for budget hardware. Yesterday, I benchmarked Unsloth Gemma 4 12B Q4_K_XL on an 8GB card. The community went wild but immediately demanded more: "Can we run a 25B+ model on budget GPUs?" Today, I’m delivering exactly that. I am running a massive 26B parameter Mixture of Experts (MoE) model locally on a standard 8GB VRAM setup with 250k full native context!. If you own an RTX 3060, 3070, 4060, or any budget GPU with 8GB of VRAM, the local AI paradigm has completely changed. The performance metrics are astonishing: - 20 tokens/sec flat decode throughput. - Stable, flat decode speed even with massive prompts. - I threw a 60k token prompt at it, and it still clocked in at 20 TPS without dropping a single frame. # What about prefill? Yes, Time To First Token (TTFT) is slightly high when swallowing massive contexts. But with a solid 200 tokens/sec prefill speed, the wait is barely noticeable and highly usable. And this is running completely without Multi Token Prediction (MTP) active. How is this possible? It’s the magic of Google's new QAT (Quantization Aware Training) quants for Gemma 4. The model weight file (unsloth gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf) is only 13.2 GB, making it the ultimate local powerhouse. # The Test Setup: CPU: Intel Core i7 RAM: 16GB System RAM GPU: NVIDIA GeForce RTX 4060 Laptop GPU (8GB VRAM) # The Secret Sauce (The -cmoe Flag) To make this work properly on any 8GB card, you must use the -cmoe (CPU MoE) flag in llama.cpp. This flag isolates the heavy MoE expert weights directly to system memory (CPU/RAM) while letting your GPU focus strictly on the Attention layers and the KV Cache. It prevents VRAM spillage and holds the throughput rock solid. # The flags: -m "gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf" -cmoe -c 248000 -v Once running, just open the UI on localhost and toggle the new reasoning lightbulb icon in the text input box to watch the model perform multi step thinking. Are you still running smaller models, or are you ready to scale up your budget local setups? Let's discuss in the repliesshow more

Christine Lagarde just said “𝗣𝗟𝗔𝗡 𝗕”… 1/21/26 Mr Pool... “𝗣𝗟𝗔𝗡 𝗕 𝗕𝗘𝗚𝗜𝗡𝗦 𝗝𝗔𝗡𝗨𝗔𝗥𝗬” 12/18 “This is a wake-up call. A bigger one than we ever had, and I think that Europe is going to look at its strength, look at its weaknesses—do you know, big SWOT analysis—and decide what do we need to be strong by ourselves? Be more independent, rely on the internal trade that we do with each other so that we can just not ignore but at least be prepared with another *** 𝗽𝗹𝗮𝗻 𝗕*** just in case the normal relationship is not restored” MR POOL CONFIRMED ONCE AGAINshow more

JENSEN HUANG UNVEILED A BOARD THAT RUNS 1 TRILLION... PARAMETER AI MODELS. THE $249 NVIDIA BOX UNDER YOUR DESK KILLS A $200/MONTH AI BILL FOR $5 IN ELECTRICITY jensen held it up on stage with one hand and called it the architecture that runs the future of ai. that same technology now ships in a $249 box smaller than your wallet the jetson orin nano super pulls 7-25 watts and does 67 trillion ai operations per second. llama 3, mistral and deepseek run locally with no api fees and no data leaving your machine most developers pay $2,400 a year across chatgpt, openai api, claude pro and cursor. the jetson costs $314 in year one and $60 a year after. 2 year savings hit $4,431 install ollama with one command, change one line of code to point at localhost, and every tool built for openai works identically. zero rewrites, zero rate limits cloud subscriptions keep getting more expensive and rate limits keep getting tighter. the people who own the box in 2026 are going to look very far ahead in 2028 bookmark this and read the article belowshow more

Right now, you may not have access to models... like GPT‑5.6 Sol, GPT‑4.6 Terra, GPT‑5.6 Luna, Claude Mythos 5, or Claude Fable 5. But you can run something surprisingly powerful today, locally, and completely free. in the next 10 mins on your 8 GB VRAM gaming laptop. Gemma 4 26B A4B QAT (MoE) delivers strong performance on a standard 8 GB VRAM GPU using Ollama, with no API, no usage limits, and no external dependencies. Out of the box, it reaches around 20 tokens per second without any optimizations. Only one command in your terminal: Ollama run gemma4:26b This means: Full offline capability (privacy by default) Zero recurring cost Competitive performance for many real world tasks Fast enough for interactive use on cheap consumer hardware If you're waiting for cutting edge cloud models, you're missing what is already practical today: a capable, local LLM that runs entirely on your own machine.show more

For a long time, staking usually meant one thing:... Lock your tokens and wait 🤷‍♂️ But some newer models are trying to connect staking with actual network activity. Wanchain 's xWAN system is an interesting example of that 👇 👉 Convert WAN into xWAN 👉 Stake xWAN 👉 Become part of the bridge ecosystem What caught my attention is that the idea isn't only about holding tokens. The bridge is being used every day as assets move between different chains, and staking is tied to that broader activity. As more users move funds across networks, infrastructure becomes increasingly important 🔥 Because today's crypto user rarely stays in one place: 👉 A transaction starts on one chain 👉 Continues on another 👉 Ends somewhere completely different Cross-chain movement is slowly becoming a normal part of crypto. And when that happens, the infrastructure behind those transfers matters more than ever. It's interesting to see staking evolving from a passive activity into something more connected to real network usage. Do you think staking models tied to actual ecosystem activity are the future? 🤔 👉show more

On this, my birthday… I thank God for a... year, though with more challenges than victories, that taught me lessons hidden in pain, in loss, in struggle. For showing me that taking a noble stand, even if it costs you, even when you’re criticized, is far better than living safe, silent, and untested. I Thank God for a year that built grit in so many of us and taught us more about the invincible power of not giving up. For the humbling truth that no one person can save the world but we all can serve in it - especially in small acts. Daily kindness. And maybe that’s how the world heals, not all at once, but a million times over. Thank you for reminding me, in a long year, to hold onto wonder, joy and even fun - as defiance, as resistance, as restoration in the long fight. And this year… thank you for a love that changed my life, for a wife who brought with her a universe of joy, and a partner to walk with, day by day, turning the ordinary into something sacred. Another year older. Another year grateful. Another year still learning how to love, to serve, to stand. (Thank you jin_sung_sa for sharing this precious birthday greeting with your followers and me).show more

🌟 [#WEVERSE UPDATE] 250811 #LEEHAN Sometimes innocent, sometimes mature,... that’s the kind of person you are. It’s a period filled with many thoughts and worries, but the way you quietly persevere is impressive, and in a way, it’s also endearing. I sincerely wish you a happy birthday, a day that will shine even more brilliantly from now on. As your teammate and friend, please never forget that no matter the hardships or sorrows, there are many people by your side you can lean on anytime. I hope we keep running forward together as hard as we do now. Let’s go together until our music echoes all over the country. I love you 🥰 Look at you moving forward steadily!show more

🚨 Grass Season 2 Airdrop – How to Increase... Your Diamonds 😱 Many people are asking: “How can I earn more Diamonds?” Here’s a clear and simple explanation 👇 1️⃣ Keep your node running as much as possible — on Mobile, Desktop, and Extension 2️⃣ Your Diamonds mainly depend on network activity (connections) 👉 The more traffic and requests your internet handles through the Grass network, the more Diamonds you earn 👉 If your connection has low activity or fewer requests, your Diamond rewards will be lower 💡 Tip: Try to maximize uptime and improve network activity. In some cases, changing your location can also help increase connections. Diamonds will play a key role in Season 2 airdrop distribution — don’t ignore this! 👇 How many Diamonds do you currently have? Let us know in the comments! Grass #Airdrop #Web3 #CryptoRewards #PassiveIncome #CryptoCommunity #grassseason2airdrop #grassdiamonds #drassairdropupdateshow more

With how amazing Tesla FSD has become, I’m at... the point now where I actually prefer to drive rather than fly in most cases. FSD allows you to travel cover an incredible amount of distance without any major fatigue since the car handles 100% of the driving, charging, and parking. It also makes the trips so much more fun and rewarding. People always ask, don’t you get bored? My answer is actually no, we get to explore so many new places along the way, take in the scenery more, and it’s just overall a more pleasant experience than manually driving and stressing about which exit to take etc. FSD handles all of that for you. If you have a Tesla, or are planning to get one, I highly suggest taking it on a road trip somewhere. It’ll be the best roadtrip you’ve ever taken.show more

It pains me to know that the belief is... that a therapist role is to "validate". I do not validate. If you are with me in a clinical setting and what you're saying doesn't make any damn sense, I'm going to have you walk me through it. What I've discovered is that a lot of the dysfunction people have is that they've never articulated their beliefs and so have no idea how absurd they are. Simply going through the exercise of having to explain yourself can be deeply rewarding. It's something we call "accountability". You literally account for yourself. Sure, I've had some female clients chimp out and flee never to be seen again at their first contact with a mirror. I've had many others weep openly that actually being held psychologically. One I will never get out of my mind was a wonderful 30-something year old woman who cried and said, "I have been in therapy for 7 years. Why has nobody ever challenged me and help me see that I'm the one causing all this shit?" That's an important question. And it's crucial that if you see a therapist, you don't see a charlatan. You can tell the good ones by their headgear.show more

Happy Father's Day, Dad. I miss you so much,... there isn't a day that goes by that I don't think of you. I keep thinking each year that it will get easier, but the truth is it doesn't. I want to show you all my accomplishments and how far l've come despite the challenges I've faced. I know you are there in spirit, and that brings me a sense of comfort. I know there are so many qualities within myself that you've given me. I'm so proud of you for the man you were. Every time I run into one of your friends I keep expecting to hear a story that changes my opinion of the angel that you were, but instead I'm even more validated each time of the humble amazing soul you were. You've protected me in life and death and that's something I'II be forever grateful for. Thank you for showing me what true love is by the way you treated my Mom. It's because of you, that I don't settle, you were her Prince Charming and I know one day I'll find mine too. ❤️🙏🏼🇲🇽show more