🤩This looks extremely useful to improve designing and debugging workflows in Claude Code. Sometimes it's hard to explain to Claude Code the exact changes you want in your app. This provides better visual cues and more direct and useful context for Claude Code. https://t.co/y7nbm3CwRV

@eunifiedworld @bytebot girl. https://t.co/Q3ODvrvnxu

i swear to god if i hear a normie say «knowledge graph» one more time… https://t.co/3MUzxTMVvD

i swear to god if i hear a normie say «knowledge graph» one more time… https://t.co/3MUzxTMVvD

excited to have the vp of at at algolia give cus a talk on how to build self improving ai systems https://t.co/6ah81HZm7L

excited to have the vp of at at algolia give cus a talk on how to build self improving ai systems https://t.co/6ah81HZm7L

Google just released FunctionGemma https://t.co/PwXVblLxnr

Nice! https://t.co/IkI5tVo2fM

Nice! https://t.co/IkI5tVo2fM

Introducing T5Gemma 2, the next generation of encoder-decoder models 🚀 Built on top of Gemma 3, we were able to build compact models at sizes of 270m-270m, 1B-1B, and 4B-4B sizes. While most models today are decoder-only, T5Gemma 2 is the first (I'm aware of) multimodal, long-context, and heavily multilingual (140 languages) encoder-decoder model out there. We hope this model enables the model research community as well as the community of devs ready to explore with new architectures. Blog: https://t.co/12ScxYcjxa Models: https://t.co/D38wNFo5Bc Paper: https://t.co/2rypSQ7Bf6

+1 more

A GGUF format model from the GLM-V series, created by the Hugging Face team. Welcome to use it🥳 https://t.co/mNAvC7NlnX

A GGUF format model from the GLM-V series, created by the Hugging Face team. Welcome to use it🥳 https://t.co/mNAvC7NlnX

GROK JUST TURNED VOICE AI INTO A REAL PRODUCT, FAST, AND EVERYWHERE xAI just opened Grok Voice to developers, and this isn’t some early experiment dressed up as a launch. It’s the same system already running inside millions of Teslas, now exposed through an API that actually works in the real world. Speed is the first thing that jumps out. Grok responds in under a second, which means conversations flow instead of stalling. That alone separates it from most voice AI people tolerate rather than enjoy. Language is the next flex. Grok handles dozens of them, switches mid-sentence, and keeps accents and tone intact. It sounds like someone talking, not software guessing. Then there’s capability. Grok can search live data, pull from X, use tools, plan routes, and control systems. In Teslas, it already acts like a co-pilot, not a novelty. No token gymnastics. Source: @xai

https://t.co/Tub46GilMJ

https://t.co/Tub46GilMJ

A decade ago SpaceX emerged from the most turbulent and challenging period in its history: surviving a major loss and doubling down with a new rocket, super-cold propellant, and a historic landing. This is that story: https://t.co/LLb0NcEfTw

"Fraud tourists" traveled to Minnesota after a friend told them state programs were "a good opportunity to make money," prosecutors say. https://t.co/dkI1RGAVOX

"Basically the government funded NGOs are a way to do things that would be illegal if they were the government." 一 Elon Musk https://t.co/XvcLhmyWSE

BREAKING: Grok 4.1 Fast just claimed the #1 spot on both the GeoMetrics and Agent Zero leaderboards. 🥇 https://t.co/JWex5lh7zb

Grok Imagine https://t.co/rVyOGio6m5

Two of Elon's most important posts in 2025. Calling out Blank Slatism along with understanding both nature and nurture have a role to play in human development is essential if we want to move past Wokeness. https://t.co/OZkFsMICfx

even sexier tbh https://t.co/xWWy6q7dxG

A generation of white people were openly racially discriminated against in the name of… Diversity. https://t.co/yFOke59CZv

¡Bienvenidos a Sora! Hoy, Sora se lanza en Argentina, Chile, Colombia, Costa Rica, República Dominicana, México, Panamá, Paraguay, Perú y Uruguay, y pronto llegará a más países. ¡Tenemos muchas ganas de ver lo que crearán! https://t.co/znb1Nm5Y7Y

LlamaParse v2 is here: simpler configuration, better performance, and up to 50% cost reduction for document parsing. We'v redesigned our approach based on user feedback to make production-ready document ingestion accessible without becoming a parsing expert. 🚀 Four simple tiers replace complex configuration modes - just pick Fast, Cost Effective, Agentic, or Agentic Plus based on your needs 💰 Significant cost improvements across all tiers, with our Agentic Plus tier offering 50% price reduction ⚡ Enhanced accuracy and reduced hallucinations, especially for complex documents with multimodal content Instead of tuning dozens of parameters and choosing between parsing modes, you now focus on outcomes while we handle the optimal model routing automatically. This means we can ship performance upgrades more frequently without breaking your workflows. Read the full announcement and learn about the new tier system: https://t.co/bsQecEK5mY

Fun article on the failure of a Claude-run vending machine in the WSJ newsroom. Reporters are amazing red teamers, creating fake policies & convincing Claude to order (and give away) Playstations & live fish. And yet… there are some hints of very viable paths forward from here https://t.co/RqILJXOcNm

+4 more

https://t.co/0RlmcWdK3U

New from @Arm + PyTorch 👇 Train smarter for the edge: • Hardware–software co-design • Extreme + QAT quantization • Mixture-of-Experts for lower latency Runs even faster on Arm with KleidiAI + ExecuTorch. 👉 Blog + 10 hrs of notebooks: https://t.co/ktI8a1UU1P #PyTorch #ExecuTorch #Arm #KleidiAI #QAT

@simon_mo_ explains why vLLM joined PyTorch Foundation, how it builds on PyTorch, and why ease of use and efficiency are central to the project. Watch here 👋https://t.co/AjLWL91i5b #PyTorch #vLLM #AIInfrastructure https://t.co/GGb3QqAmBS

@simon_mo_ explains why vLLM joined PyTorch Foundation, how it builds on PyTorch, and why ease of use and efficiency are central to the project. 👋Watch here: https://t.co/AjLWL91i5b #PyTorch #vLLM #AIInfrastructure https://t.co/oo6piIiyOq