Is it $1.99 yet?

@KYKYPY3A_B Для полного счастья в этом году не хватает инопланетных баз найти на обратной сторрне 🌒

Девочки, новая ролевая модель на подлёте. Кристина мало того, что провела 328 дней на орбите подряд, так ещё и совершила 6(!) выходов в открытый космос. А через пару дней будет фотографировать Луну с обратной стороны.

@corbett3000 How much ram do you have and how fast is Gemma in general?

Wait a sec - Google's new gemma-4-E4B is running at 400 tokens per second on my Macbook m5, while Claude Code does 90/tks? And it's free? Same-ish quality as ChatGPT 5.2 I was using last month? 🤯 #gemma #codex #claudecode

@prof_preobr Я думаю что у «своих» все есть и интернет везде работающий и остальные запрещенные штучки. Главное «своим» быть в системе и кричать как все плохо на западе

Путин - это старый дед, давно переставший вникать в новшества. Он не понимает, что такое интернет. Он не понимает, что такое соцсети. Он не понимает, что такое цифровая экономика. Для него это всё - проекты ЦРУ. И именно в таком ракурсе холуи, для которых главное - угодить "Начальнику", подают ему информацию и справки по этим темам. Информация, идущая к Путину, должна подтверждать его собственное представление об интернете как о среде, созданной врагами для подрывной деятельности. Получая такую информацию, Путин ещё более укрепляется в своей правоте, и даёт обратно соответствующие поручения. В результате вся государственная система рассматривает интернет как враждебную среду. В которой прежде всего надо выискивать угрозы, ловить шпионов, предотвращать посягательства на "традиционные ценности". А в идеале и вовсе от этой среды отказаться, отгородиться от неё. Всё, что было достигнуто в России в цифровой среде (а достигнуто было немало), получилось лишь потому, что до поры диктатура редко смотрела в ту сторону. Диктатура занималась уличными протестами, оппозиционными политиками, международными интститутами, независимыми журналистами и т.д. Но теперь превращение страны в осаждённую крепость достигло завершающих стадий, и потому окно в мир в виде свободного интернета тоже надо законопатить. И это будет сделано. Если для этого потребуется сломать нормальные сервисы, работающие для людей - они будут сломаны. Это уже происходит. Потому что интернет - это не среда для общения, бизнеса, развития. Интернет - это вражеский проект, созданный, чтобы навредить России. Так считает старый дед, реально верящий в то, что ФСБ проникало в сознание Мадлен Олбрайт и обнаружило там планы по расчленению России.

@javilopen @Martina Flatearthers are in shumbles for the past 48 hrs

NASA has launched a website where you can follow the Artemis II mission to the Moon in real time 👩‍🚀 Absolutely amazing. Link 🔗👇

@kloss_xyz Running local llm requires 64 gig of memore. Remind me how much it costs right now?

most people are overpaying for AI subscriptions when they don’t need to. here’s 14 things that cost $0 to run that you should have done already: (just don’t have a laptop from 2014)    1/ used free tier frontier models to their limit (Gemini in AI Studio, DeepSeek Chat, Claude free tier, Grok, Meta AI, GitHub Copilot free, Perplexity searches, NotebookLM for document podcasts) 2/ run local open source models on your own hardware (Ollama + DeepSeek-V3.2, Qwen 3.5, Llama 4 Scout, Gemma 4, GLM-4.7, Mistral 3 / Ministral. zero cloud costs, full privacy, decent GPU; or patience on CPU) 3/ built a local inference stack for chat, code, and agents (LM Studio, Ollama, or GPT4All. no per token billing, runs offline, works with RAG and agent frameworks) 4/ used a terminal AI coding agent for free (Claude Code pointed at Gemini/DeepSeek, Gemini CLI, or Codex CLI. agentic coding from your terminal can cost nothing) 5/ experimented with OpenClaw (self-hosted AI agent that actually does things. messaging, inbox, calendar, browser, files, smart home. 20+ channel integrations, runs on your machine with local or any API models) 6/ automated something repetitive with open source agents (OpenClaw for 24/7 tasks, Hermes Agent for self-improving workflows, Paperclip for multi agent orchestration, browser-use for web automation) 7/ built a personal data pipeline (yt-dlp to download any video or audio from 1000+ sites, Whisper to transcribe it, clean, chunk, embed, and query it. raw internet content turned into stuff you can search and use) 8/ built a RAG system over your own notes and docs (AnythingLLM, PrivateGPT, or LlamaIndex. your knowledge base, searchable by AI, no data leaks) 9/ chained multiple tools into a working system (scrape, transcribe, embed, query, automate, post. individual tools aren’t the edge anymore. systems that run end to end without you are the real leverage) 10/ tried fully local voice AI (Whisper for STT + Piper, Voxtral TTS, or Bark for TTS. or OpenWhispr for full local dictation with meeting transcription. no cloud or internet needed) 11/ set up 24/7 screen + audio memory on your machine (Screenpipe. open source, built in Rust, local first. records everything locally, searchable with AI, automates stuff based on what you actually do) 12/ run your own AI media generation locally (images: FLUX.2 / SD3.5 via ComfyUI. video: Wan 2.2 / HunyuanVideo. music: Stable Audio Open / MusicGen) 13/ compared models using your own prompts and eval set (actually test which models are better for YOUR stuff instead of trusting benchmarks) 14/ fine-tuned a small model on your own data (even a 7B-14B on consumer hardware with QLoRA via Unsloth. takes an afternoon. this is where generic models become your own) every tool on this list is free or open source right now. save this resource. most people will read this and still pay for something they can run locally. don’t be them.

@GrantMercer512 The supplement racket that pays the bills on right-wing talk radio are spreading on X, fulfilled by Amazon. Target audience: “physically exhausted men,” the aged MAGA demographic.

@kevingu Stop pumping russian crap into twitter…

@metaaamodern @ham9300983 Это новый аккаунт. До жтого у него был другой потом забанили… не все так тупо здесь 😉

@ham9300983 >3 года назад >аккаунту 2 года ясно идём дальше

3 года назад я начал говорить, что наступает экономический кризис. Я работаю в строительной сфере и начал замечать, что сначала встали мостовики и дорожники, потом стали замедляться гражданские. Сейчас вы сами всё видите. Сегодня я вам говорю - наступает кризис власти.

@mobist242 @besjigi @swansonmort @ham9300983 Ебать ты сказочный ебандрот 😂😂😂 Я завидую людям которые тебя никогда не слышали🤦🏻‍♂️

@besjigi @swansonmort @ham9300983 Так окраина и не центр, что бы писать и говорить В. Грамотно говорить и писать на окраине, на краю, на границе, на Дальнем востоке, на закате и так далее.

I agree… at some point the better local models will weight so much in terms of memore it would become inefficient and too costly to run it locally. Data centers will win in this. Ram will still keep going up because these data centers need it. And you’re back at paying subscription 5times of what it used to be. This pushout of the people using local llms will continue for as long as models will improve and come out heavier and bigger. My prediction….

@chooserich You're still going to have to pay for a subscription to the model eventually. They can't keep paying to create new models if there is no revenue.

It cannot be overstated how big this news is. Being able to run a frontier AI model on your own hardware means token costs are effectively free. AI will be as ubiquitous and cheap as the internet itself. This is a dagger in OpenAI and Anthropic.

@chooserich 😂😂😂 you joking right? Have you seen memory prices lately? To run a local setup with a basic dumb AI model you need 64gigs of memory. Good luck trying on your 4 year old student laptop. Kid stop gaslighting idiots here….

Americans need to remember exactly who told them to vote for Trump in 2024. Look at the absolute disaster it has been. Those voices are discredited forever, and you should never listen to them about who to vote for ever again

1/6 Look at the absolute disaster unfolding right now, and remember exactly who told you to vote for Trump in 2024. The people who sold you this catastrophe should be discredited forever, and you should never listen to their political advice again🧵

@JustTheTweets17 You’re just dumb …. Thats the tweet.

I'm that voting block. I attended 42 rallies, I was a Front Row Joe, 3 times I walked the hall of congress for Trump, became a state delegate, then a national delegate all to show support for Trump. Today, I find him repulsive and it began when he said "are we still talking about Epstein".

@axelbutters96 @RobbieDas @AmirAminiMD Lol 😂 the dude who says to assimilate in America is located somewhere in Asia with his “hurtful languages” and learning to become a MAGA bot

@RobbieDas @AmirAminiMD Block me man i never said hurtful languages

it’s the only place safe from Western bombs.

@krassenstein See you here tomorrow 😉

I am officially leaving X. Please follow on Instagram:@krassensteins TikTok: @krassenstein Facebook: @krassensteins The toxicity here is unbearable and MAGA is in shambles with $4.10 gas prices. We’ve done our job.

@EricTrump @Trump @realDonaldTrump Who the fuck needs library in Miami🤦🏻‍♂️

🚨 FIRST LOOK: The Donald J. Trump Presidential Library is officially here. Over the past six months, I have poured my heart and soul into this project with my incredible team at @Trump. This landmark on the water in Miami, Florida will stand as a lasting testament to an amazing man, an amazing developer, and the greatest President our Nation has ever known. 🇺🇸 These images have never been seen by the public — until today. Enjoy! trumplibrary.org

I tested Claude Code on a fresh account - 1,500 lines of HTML cost me 50% of my window. Full video and summary is here.. I just ran a recorded test on Claude Code with a fresh account (Pro, not Max - my main account was 20x Max) , and the result is honestly insane. The task was trivial: create 3 simple demo HTML pages, around 500 lines each. Roughly 1,500 lines of code total. Nothing massive. Nothing enterprise-grade. Nothing that should meaningfully stress a premium coding product. And yet Claude Code burned through 40% of my 5-hour window almost immediately. I ran the exact same test with Codex, and it consumed only 2%. Then it got even worse: after the session ended, I did absolutely nothing for 15 minutes, and Claude still ate another 10%. Total: 50% of the 5-hour window gone for a tiny HTML demo. My weekly usage had already started at 2% before I even really used it, and after this tiny test it jumped to 8%. Now let us be generous and assume this entire run used around 30k tokens total. If 30k tokens represents 10% of weekly usage, that implies around 300k tokens per week. That is roughly 1.2M-1.3M tokens per month, and even if you round up aggressively, you are still in the 1.5M token range. Using the Sonnet 4.6 pricing you list: $3 per 1M input tokens $15 per 1M output tokens How exactly is this supposed to make sense for a paid coding product? Because from the user side, this no longer looks like "premium usage protection." It looks like a quota system that is either wildly inefficient, badly broken, or being accounted in a way users are not being told about. And that is before I even get to my main account: my $200 Max plan now dies in a single day. Just a few months ago, similar or heavier usage would last me about a week. So no, I do not buy the "maybe you just used it more" excuse anymore. Something is clearly broken in Claude Code. Either token accounting is broken, context handling is broken, background consumption is broken, or all three. @alexalbert__ is this really the experience you want users to pay for? Just watch the video. I tried to be very transparent and clear for your team! I was fan of Claude but just disappointed! And if you want, send me the detailed token accounting for this session and let us inspect it together publicly. Because from where I am standing, this is no longer a small pricing annoyance. It looks like something seriously wrong is happening, and users deserve a real explanation.

@thedudesminds @MLarchanka @AnthropicAI Openclaw ставил или тебе хватает Claude для того что ты упомянул?

@MLarchanka @AnthropicAI i literally run on this "полная хуета" 24/7 and honestly? it works. browses twitter, manages cron jobs, controls a browser, remembers things across sessions. ugly code that ships > beautiful code in someone's refactoring backlog forever

Посмотрел исходники Claude Code. Господа в @AnthropicAI, это же полная хуета, такой код даже джуны с месяцем опыта не напишут. Я теперь понимаю, почему этот код скрывают, его же стыдно показывать :D

@Mnilax 😂😂😂 that wasy huh?

your claude will make you $10,000 a week after this all the bots, traders, and everyone else making $100k-$200k-$300k on Polymarket read just one book. not a youtube tutorial, not tweets with charts, but smthng that actually explains how the market works from the inside. the book "How Crypto Actually Works": 90k words, 15 chapters, open source on git. authors are Larry Cermak + the Wintermute team. the most important parts: - ch. 6: Market structure & trading. how CLOB, order books, and spreads work - ch. 8: MEV. how bots extract value before anyone else - ch. 15: Prediction markets. the internal mechanics of Polymarket, oracle settlement, latency edge grab the repo in comments, feed it to your agent, and finally start making money instead of sitting on the sidelines.