Sam Altman’s AI paradox: Warning of a bubble while raising trillions https://t.co/QhV0JmUSuv @fortunemagazine

Strategy Tumbles to 4-Month Low as Crypto Stocks, Digital Asset Treasuries Sink https://t.co/lfM0kLBE94 @coindesk @sndr_krisztian

Strategy Tumbles to 4-Month Low as Crypto Stocks, Digital Asset Treasuries Sink https://t.co/lfM0kLBE94 @coindesk @sndr_krisztian

Californians Say AI Is Moving 'Too Fast' https://t.co/Ors0f9NwGt @billyperrigo @time

Californians Say AI Is Moving 'Too Fast' https://t.co/Ors0f9NwGt @billyperrigo @time

How AI-enhanced hackers are stealing billions https://t.co/ZVaPsJSYFp @theeconomist

How AI-enhanced hackers are stealing billions https://t.co/ZVaPsJSYFp @theeconomist

Insightful Pulse of Fintech – UK perspective https://t.co/tSHmQZJUAH @kpmguk @kpmg https://t.co/IyOwS6th8Y

Raising $4.7 million in a week: What one founder learned after moving his AI startup from Germany to San Francisco https://t.co/wTrKajCPdE @kelseyvlamis @businessinsider

Thelonious Monk turned wrong notes into the right ones. https://t.co/YfZLuv5y3q

Thelonious Monk turned wrong notes into the right ones. https://t.co/YfZLuv5y3q

Link to FAQ https://t.co/cFjEhuE35U

How do you write evals to test your AI's ability to express uncertainty or refrain from answering when it shouldn't? Answer: assemble the right dataset and use it as a test harness + look at the data 😅 . Links in reply https://t.co/b678c2vsBf

@strickvl Didn't have much in it https://t.co/vx6DLt4Xq1

Made a small video about justfiles. They are just so darn amazing. https://t.co/jInT5sndrP

Made a small video about justfiles. They are just so darn amazing. https://t.co/jInT5sndrP

After a million years I finally uploaded a new video. This time about RoPE! Let me know what you think! https://t.co/DchLMHtSmW

I took a RoboTaxi the other day and noticed a super cool 'Find My Vehicle' feature. Also, super cool that I can use my phone to unlock the car, open the trunk, flash the beams, and honk the horn. Super easy to find the car anywhere. @Tesla @Tesla_AI @elonmusk https://t.co/6BjrNkefal

Huge 3D Printing Breakthrough! Caltech is printing metal atom by atom. A hydrogel scaffold soaked in metal salts becomes the blueprint, the gel burns away, and oxygen is stripped in a hydrogen-rich chamber, leaving alloys crafted with surgical precision. They can change metal ratios on the fly. Flip from a copper-heavy mix to a Cu12Ni88 alloy nearly four times stronger, without altering shape or tooling. Every property, from strength to density, is now designed, not guessed. This is metallurgy reinvented. Also a game changer for Aerospace applications. https://t.co/Wcr2mNaKgb

https://t.co/X95TESL90l

https://t.co/X95TESL90l

@JacobB1290H umm update the new call screening feature in iOS26 does not seem to work at all? my phone buzzed alive with one of them just now: https://t.co/Cb3UL1M8WZ

Omg new @Explosions_Fire https://t.co/xVR29m2R1Y

ARC-AGI-3 Preview - 30-Day Learnings 30 days ago we released a preview of our first Interactive Reasoning Benchmark Our goal was to ship quick, learn from the community, and inform the next >100 games. Here’s what we learned after 100s of agents and >3,900 game plays: https://t.co/o07wrUBzWW

Has GPT-5 Achieved Spatial Intelligence? GPT-5 sets SoTA but not human‑level spatial intelligence. My notes below: https://t.co/taQms63ji3

This doesn't seem good does it? https://t.co/9kAe4Oe9tg

@sbeastwindy @danielhanchen It's pretty simple to just change 2 lines in the notebook and verify the findings independently. https://t.co/xlzpTIASDw

Price Alerts for Indian stocks are now available to all Perplexity Pro and Max users in India! Currently supported on web and mobile web. Mobile app support and availability for free users coming shortly! 🇮🇳 https://t.co/GHC3k3C6Bm

ComputerRL: Scaling End-to-End Online Reinforcement Learning for Computer Use Agents "To support scalable and robust training, we develop a distributed RL infrastructure capable of orchestrating thousands of parallel virtual desktop environments to accelerate large-scale online RL. Furthermore, we propose Entropulse, a training strategy that alternates reinforcement learning with supervised fine-tuning, effectively mitigating entropy collapse during extended training runs."

NovoMolGen: Rethinking Molecular Language Model Pretraining "there remains limited understanding regarding how standard language modeling practices such as textual representations, tokenization strategies, model size, and dataset scale impact molecular generation performance. In this work, we systematically investigate these critical aspects by introducing NovoMolGen, a family of transformer-based foundation models pretrained on 1.5 billion molecules for de-novo molecule generation. Through extensive empirical analyses, we identify a weak correlation between performance metrics measured during pretraining and actual downstream performance, revealing important distinctions between molecular and general NLP training dynamics. NovoMolGen establishes new state-of-the-art results, substantially outperforming prior Mol-LLMs and specialized generative models in both unconstrained and goal-directed molecular generation tasks"

huggingface: https://t.co/OU4qx9S4cg abs: https://t.co/CUpqmuo378

Exploring Efficiency Frontiers of Thinking Budget in Medical Reasoning: Scaling Laws between Computational Resources and Reasoning Quality "This study presents the first comprehensive evaluation of thinking budget mechanisms in medical reasoning tasks, revealing fundamental scaling laws between computational resources and reasoning quality. We systematically evaluated two major model families, Qwen3 (1.7B to 235B parameters) and DeepSeek-R1 (1.5B to 70B parameters), across 15 medical datasets spanning diverse specialties and difficulty levels." "Domain-specific patterns emerge clearly, with neurology and gastroenterology requiring significantly deeper reasoning processes than cardiovascular or respiratory medicine."