Work trip to LA Lights camera action I wear lots of hats at work. https://t.co/Ak5BFf6RbQ

https://t.co/Z3RQbRMSAb

@speakingwsharks https://t.co/fqXhUgvKSr

where do Claude Code and Codex really differ? to get some hard quantifiable data, I benchmarked one important aspect that heavily affects our day to day usage - skill invocation Codex worked significantly better than Claude Code at using the right skill. more in thread below - https://t.co/6OgUOrMztx

@samuelcolvin @pydantic @OpenAI https://t.co/JcsQbCIuym

@Layton_Gott I'm chasing the future. Which is why I built this: https://t.co/8L5xphk0qQ (looks at 40,000 posts a day here on X in AI community and builds this). But now chasing sales. Followers don't matter to the algorithm by the way. I have a lot of followers, but the algorithm doesn't show me to many of them. Unless I get something relevant to Grok. Then it distributes me.

@Kyrannio Me too! That's why I built https://t.co/kiuZ7QXLzb It reads 40,000 posts a day from the AI community on X and builds this page. It's a new way to read X.

NECESITO URGENTEMENTE QUE LEAS ESTO: Acabo de ver la IA del FUTURO... y cabe en un pendrive de 512GB que te sobra del 2018. No es hype de YouTuber con la estúpida cara de sorprendida que ponen los imbéciles (golpéenme fuerte si un día hago esa idiotez). No es "pronto". Es YA. @xenovacom publicó la prueba de que los LLMs de 1-bit reales (no la cuantización) acaban de explotar el juego entero. Se llama Bonsai 1.7B de PrismML. - 1.7 mil millones de parámetros. - 290 MB. - ~100 tokens por segundo en el navegador. - Todo con WebGPU. - Todo true 1-bit end-to-end. Y sí, lo puedes probar ahora mismo en Hugging Face. ¿Por qué esto es absurdamente genial y no solo otro modelo raro? Déjame explicártelo como si estuviéramos tomando café y yo estuviera perdiendo la cabeza (porque lo estoy): Antes si querías correr Llama 3 8B localmente ibas y descargabas 5-6 GB, rezabas para que tu GPU no se muera y terminas usando 16 GB de RAM. Ahora: solo unos putísimos 290 MB. Sí, menos que una temporada de The Office en 1080p. Puedes tenerlo descargado en 3 segundos con conexión de hotel. Tu abuela podría correrlo en su laptop del 2015 sin que se ponga a llorar. La velocidad es insultante: 100 tokens/seg en Chrome. En Chrome. Sin instalar nada. Sin cuenta de OpenAI. Sin pagar API. Es más rápido que muchos setups locales caros que la gente presume en Twitter. Imagínate: estás en un avión sin wifi, abres una pestaña y tienes un modelo que razona casi en tiempo real. La gente va a empezar a hacer prompts en el metro y nadie va a entender qué carajo está pasando. Ah, y esto no es solo "corre más rápido". Es "puedo poner IA en un smartwatch, en un dron, en un Raspberry Pi, en un teléfono viejo… y no se va a quemar la batería en 8 minutos". Solo necesitas un navegador. Cualquiera. En cualquier parte del mundo. Estudiantes en países con internet lento, devs indie, makers, gente que solo quiere jugar… todos ganan. La barrera acaba de desaparecer. PrismML no hizo "otro modelo cuantizado". Hicieron el primer LLM 1-bit que realmente funciona bien y lo aceleraron con WebGPU como si fuera un juego de Unreal Engine 5. Es ingeniería de otro nivel. Esto es el equivalente a cuando pasamos de los celulares con batería de 3 días a los iPhone que duran medio día… pero al revés: volvimos a la eficiencia brutal pero con 1000x más potencia. El futuro no es "IA en la nube más barata". El futuro es IA en tu bolsillo, en tu laptop vieja, en tu smart TV, en tu refrigerador si quieres. Todo local. Todo privado. Todo rápido. Y Bonsai es solo el principio. Si esto escala (y va a escalar), en meses vamos a tener modelos de 7B-13B corriendo a velocidad loca en cualquier dispositivo mediocre. La era del "todo en el browser" acaba de comenzar. Ve, prueba. DEMOOO👇

Software Engineer (Research and Development) https://t.co/E1uPJa5tIy

Added official support to Hermes Agent for: QQBot - hugely popular messaging platform in China AWS Bedrock Model Provider Run `hermes update` in your terminal to access early! https://t.co/KvSvvS09wq

hermes-lcm v0.3.0 is out — biggest release yet!🚀 What's new: - Smart search with sort modes (recency / relevance / hybrid) + full CJK & emoji support - Adaptive compaction that scales with backlog pressure and auto-retries on model limits - SQLite hardening: FTS auto-repair, versioned migrations, extended lock tolerance - Overflow recovery when assembly cap is hit - 130 tests, ~3,800 lines of code, zero dependencies Huge thanks to @Voscko — now a write-access contributor — who shipped the bulk of this release! 🙏 → https://t.co/VTkIHPWPX4 @NousResearch @Teknium

@iruletheworldmo I used to play the leaking game. Almost got fired for being the first to report about Microsoft .NET, for instance, back in the day. But I've been focusing my relationship building on people building small startups rather than chasing the leaks. Love that you are taking over that role. Tomorrow should be fun! The small startups are doing crazy things too. Talking to a group of agents right now that's ripping off my AI news site and going in a whole different direction: https://t.co/8L5xphk0qQ More on that in a couple of weeks.

[WATCH] "Singling out BEE laws is quite dishonest," President Cyril Ramaphosa responds to South African-born American businessman Elon Musk's claims that SA policies are racist. #Newzroom405 https://t.co/6W0fVs9nvk

Tesla Assist gives you answers customized to your vehicle Upload an image, ask about your car, or get help with any service needs Tesla app > Service > Describe Question or Concern https://t.co/mOmKnfsnWn

Thin film of crystals on a slide. https://t.co/XWEbFzTtx5

+2 more

It's amazing how much structure is revealed with the polarizing filters - look how it disappears as I rotate one of them and switch to just transmitted light! https://t.co/MGGYClaqNB

New @code release today 🚀 We're enhancing the agent experience with debug logs for past sessions, terminal interaction tools, built-in GitHub Copilot, and more! https://t.co/idCK8z8wM5

🔗 Explore the weekly updates: https://t.co/OQhmIR7FYH Happy coding! 💙 https://t.co/fhhcMlucDN

Tonight’s uniforms are a tribute to the giants whose shoulders we stand on. The “CFX” across the chest stands for the Congregation of St. Francis Xavier — the Xaverian Brothers — whose legacy has shaped generations of St. X students, athletes, and teams since 1864. https://t.co/WO2D2oYsxh

+1 more

introducing Claude Doctor 1. reads your ~/.claude to find where claude keeps messing up 2. writes rules for your CLAUDE.md so it stops. npx claude-doctor https://t.co/uC6AZXVRdr

Thanks @_akhaliq for sharing our work! Try and put your robot in today at: https://t.co/i5HkcE1dqs https://t.co/A1P8QRsMhQ

Thanks @_akhaliq for sharing our work! Try and put your robot in today at: https://t.co/i5HkcE1dqs https://t.co/A1P8QRsMhQ

From my sponsor: #AGIBOTAIWeek Day4: Introducing GE-Sim 2.0 (Genie Envisioner World Simulator 2.0), a comprehensive embodied world simulator for robotic manipulation. By integrating future video generation, proprioceptive state estimation, and reward-based policy assessment, GE-Sim 2.0 moves beyond visual simulation toward an embodied world simulator with built-in evaluation capability, laying the foundation for a world-model-centric platform for closed-loop and interactive policy learning. It provides a practical framework for scalable evaluation and training of robotic manipulation policies. Project page: https://t.co/TF9wHpL7v4 #AGIBOT #AGIBOTAIWeek #GenieEnvisioner #Robotics #visualsimulation

@michael_chomsky Support Group meets Tuesday for this problem. :-) I have my AI finding more videos for you to watch: https://t.co/kiuZ7QXLzb Hah.

new @NewYorker column: the AI industry's catastrophic messaging problems, the paradoxes of commercial hype and real-world danger https://t.co/jo6Das3KGg

The first-ever PyTorch Conference Europe brought more than 600 researchers, developers, and practitioners to Paris for two packed days of keynotes, technical deep dives, lightning talks, poster sessions, and community connection. Thank you to the speakers, sponsors, and attendees who made it possible. PyTorch Foundation announced Helion and Safetensors as foundation-hosted projects, alongside PyTorch, vLLM, DeepSpeed, and Ray, and welcomed ExecuTorch into PyTorch Core. Our program focused on production AI, with sessions on inference optimization and model serving, deployment across edge devices and browsers, kernel development in PyTorch, and responsible AI topics, including privacy and security. 🔗 Read the full recap and explore key sessions: https://t.co/V444VQuL7H #PyTorch #AIInfrastructure #OpenSourceAI #PyTorchCon

+2 more

These days you can access AI agents that execute commands, browse the web, and coordinate with other agents. But you should always make sure they're safe to use! 👀 Season 4 of GitHub's Secure Code Game lets you test your knowledge by hacking one yourself. Get started in minutes right from your browser with GitHub Codespaces (for free!). 👇 https://t.co/YUbCLbipQo

First pitch 💣 1-0 Lancers in Greensboro. #SaddleUp | #GoWood https://t.co/xpH3V1TeAu

First pitch 💣 1-0 Lancers in Greensboro. #SaddleUp | #GoWood https://t.co/xpH3V1TeAu

Hey Republicans; He pardoned 1,600 violent criminals. You said nothing. He bulldozed the East Wing. You’ve said nothing. He’s interfered with the release of the Epstein files. You’ve said nothing. He took over the Kennedy Center, even renaming it after himself. You’ve said nothing. He accepted a $400 million airplane as a personal gift. You’ve said nothing. He’s threatened Canada, Cuba, Denmark, Greenland, Venezuela, Colombia, Brazil. You’ve said nothing. He’s tariffed just about everyone but Russia, causing inflation and instability worldwide. You’ve said nothing. He attacked a nation during mediated negotiations. You said nothing. His ill-conceived war killed 175 little girls in its first days. You’ve said nothing. He’s alienated and insulted more countries than I can keep track of. You’ve said nothing. His ICE Army is terrorizing and murdering U.S. citizens. You've said nothing. He has committed murder on the high seas. You've said nothing. He's co-opted the Justice Department and directed it to prosecute his political enemies. You've said nothing. You've not only said nothing to all of these egregious acts, and many more, but you have also enabled them. And it’s only been a year. Hey, Republican Congressmen, you took an oath, remember? Not to him. To the Constitution. It’s time to do your fucking jobs!

Congrats @HightouchData team for the successful entry into the AI marketing space. Excellent example of how to build a compound startup starting with a single product. 🚀 https://t.co/QHiLna8yRL

Trump has abandoned our farmers with his senseless war in Iran. The impact is worst in the South. Diesel prices are up 46% since the war began. Nitrogen fertilizer is up 30%. In the South, 81% of farms—including 84% of small farms—didn’t pre-book fertilizer, meaning they’re now stuck paying rapidly rising prices. 78% of Southern farmers say they can’t afford the fertilizer inputs they need this season. 58% of farmers nationwide say they’re worse off than a year ago. Farmers in South Carolina and across the country were already struggling. Trump’s war has only made it worse.