I asked Claude about the air conditioning debate in Europe, and it really didn’t pull any punches. https://t.co/bA29683Hgh

@elonmusk Using their logic, you can argue USAID caused 9/11 then murdered 4.7 million people after. https://t.co/XZ28E0Zjsy

USAID: Using the methodology the Lancet used to determine cuts to USAID's funding resulted in four million deaths, @GKMasterson determined that USAID funding itself resulted in 187 million deaths. Either the Lancet's methodology is bogus or USAID has killed more humans than the US military has in our nation's history.

Stephen Miller on USAID: "There is probably no entity in the government that is more of an entrenched power center of unelected bureaucrats with less accountability, less oversight than USAID. They funnel money to their cronies all over the world through a swampy network of NGOs and cutouts, including some very troubling elements of focusing on things like regime change and destabilizing foreign countries."

.@RandPaul Asks Samantha Power: 'Did USAID Fund Coronavirus Research In Wuhan China?' "Should we be funding the Academy of Military Medical Research in China? ...Some of the research proposals in 2018 were the Wuhan Insitute of Virology asking for money to create a virus with a furin cleavage site. A SARS-like coronavirus with a furin cleavage site. That's exactly what COVID-19 turned out to be. So we want to know if there were other research proposals you either granted or denied that were in the same vain as creating viruses that could have become COVID-19. We can't tell because you won't give us information... I now have 25 Senators that have sent you a letter, and you aren't responding... You are being dishonest."

The weekend before, we're sponsoring the official @cerebral_valley hackathon on June 27. Come build something epic: https://t.co/c0CottRG0q

AI agents are easy to demo. Production is a different problem. Enter CUGA. 🦉 An open-source agent harness that lets you focus on building instead of plumbing: https://t.co/5jRzs87mOV https://t.co/2RkUjdyllT

Introducing Claude Tag, a new way for teams to work with Claude. In Slack, Claude joins as a team member with access to the channels and tools you choose. Tag Claude in and delegate tasks to it while you focus on other work. https://t.co/R2C6A5Kcye

Tomorrow @sh_reya and I kick off this free AI product engineering mini-course. Topics covered over 12 talks: 1. Design/UX & Evals 2. Retrieval 3. When & how to use open models effectively With these legends: @TheZachMueller @bclavie @xeophon @GoAbiAryan @barrowjoseph @willccbb @marek_galovic @beirmug Sign up & learn more here https://t.co/ssygeCw8eT

Another riddle for @RoKhanna. Why did the President of Mexico agree with DOGE and celebrate the closure of USAID? Are you accusing her of supporting mass child murder too? https://t.co/5pzwP8El3Y

@ctatedev Love it a lot. Having done agent evals for a couple of years now, this feels just about right. Just shipped an easy way to try out this and some of the cool stuff with the Eve framework. https://t.co/M2oYmpd5tb

@jxnlco https://t.co/SH3PPZgFfU

@jxnlco https://t.co/SH3PPZgFfU

@gatorgar Did someone ask for a study to prove this? https://t.co/IqbQPS2TSX https://t.co/kzW63BR1M2

Mistral claims SOTA performance on OlmOCRBench, a popular optical character recognition benchmark, but that isn't the case. We have a public leaderboard on @huggingface, where Mistral OCR 4 currently ranks #3, behind open models like Chandra OCR 2 by @datalabto https://t.co/aQOrJzfZ2K

you can download them now from Huggingface: - https://t.co/KcCsO5EiVq - https://t.co/8YBTF8MI6J

“We’re here to fuck all the white girls and fuck the government. We’re over here to breed, we are going to take over.” That’s what one of the Dewsbury grooming gang rapists told his victim. Not poverty. Not “cultural misunderstanding”. Not random crime by “Asians”. This was muslim men operating with a clear religious and racial ideology. White non-Muslim girls were “slags”, “trash”, and “kuffar whores”, fair game for grooming, gang rape, trafficking, and beatings. Muslim girls were “pure”. White girls deserved it. Some perpetrators quoted scripture and framed their actions as religiously justified superiority over infidels. When the government repeatedly sides with the ideological rapists over its own children, when it betrays fathers and communities trying to protect their daughters, don’t act shocked when people lose faith in the system and take matters into their own hands. This is demographic and cultural conquest. An existential attack on Britain enabled by the very people sworn to defend it. The absolutely filthy anti-white, anti-Christian racist Mehdi Hasan calls the Casey report and the whole push for truth “bullshit.” The rapists’ own words in court and the girls they destroyed say otherwise. Denial this loud only confirms who’s still protecting the narrative and why.

.@DAGToddBlanche announces charges against 455 defendants across 45 states and territories for healthcare fraud schemes involving over $6.5 billion in false claims submitted to Medicare, Medicaid, and other programs. https://t.co/ofeHhUAbmX

The rape gangs were allowed to flourish so rapidly partly because it was working class communities affected, and the establishment was happy to look the other way. The rape of English towns and cities. Tolerated, enabled, allowed and encouraged. The cancer must be eradicated. https://t.co/41hGLSF3cp

A Syrian national was charged with diverting millions in USAID funding to support a terrorist organization https://t.co/7SvDDzfX5t

I'm digging the eve agentic framework from Vercel. I like that everything is files, from the tools to the skills to the evals. More importantly, it's gets you building with agents fast. Very promising. If you like TypeScript you will dig this too. Get started with eve ↓ https://t.co/0dQEJMjack

While SGLang provided Day-0 support for DeepSeek-V4, the collaboration between the @lmsysorg and @NVIDIAAI engineering teams has taken its production performance to the next level. According to the public SemiAnalysis InferenceX dashboard, the GB300 disaggregated lane (DeepSeek-V4 Pro, FP4, 8K/1K) saw a 5x throughput increase—surging from ~2,200 to ~11,200 tok/s/GPU at identical interactivity levels. These updates sustain high throughput much deeper into target interactivity ranges most deployments target, while also driving a 2.9x lift on the Blackwell Ultra aggregated lane. Find the full technical breakdown in the comments below:

Hey @claudeai Opus 4.8 let's build a fully procedural spider in @threejs🕷️ …so we did. Feet-driven IK + a Cruse-rule gait = it walks any terrain. Then we built a 42-scenario test harness and drove the locomotion to 100%. https://t.co/5HU9BBvpdf

Create your first AI-native App with natural language meet KroWork 0.9.X one month, with 5 new versions shipped, here is what’s new: 1. Intelligent Apps: software with built-in AI 2. Share the app you build with One-Click 3. Multimodal capability & latest models support Your best AI workflows should become real desktop apps, with NO tokens burnt on reruns Some real cases below to show you how it works:

DataClaw0: agentic tailoring for raw multimodal streams Make any data you want. A 9B model filters noise from videos, GUI, and embodied data. It reorganizes signal into dense supervision using factual anchors and semantic synthesis. Five domains. SFT plus GRPO. Benchmark included

World Action Models: A Survey https://t.co/CgpeYgTH8c

paper: https://t.co/RuxE0wo0W6

PlanBench-XL Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool Ecosystems https://t.co/MRaAlMS4Id

paper: https://t.co/yvlfPuYCQn

*taps sign https://t.co/ayNJdwCTOf