Are scaling laws finally working for time series foundation models? Today, @datadoghq is releasing Toto 2.0 weights in Apache 2.0 on @huggingface. It's a family of open-weights TSFMs from 4M to 2.5B parameters, where every size beats the last from a single hyperparameter config. First across the leading benchmarks: BOOM, GIFT-Eval, and TIME. Most TSFM families ship multiple sizes that all perform roughly the same. This one doesn't. Why it matters: scaling laws gave language and vision a predictable relationship between compute, data, parameters, and downstream performance. Time series hasn't had that curve until now. Once you have it, you can scale data and compute with confidence, and start asking which new capabilities emerge at the next order of magnitude. 2.5B open-source weights: https://t.co/prpcGoCw0U 4M open-source weights: https://t.co/5d6rw5NYL2 Blogpost: https://t.co/xKazgTMh1I

+2 more

This one blew my mind back in the day! From "Learning Invariant Features through Topographic Filter Maps" (2009) by @koraykv, @MarcRanzato, @rob_fergus and @ylecun https://t.co/IjzRi4R9X6

We're releasing a whole new category of voice models. Introducing DramaBox — our state-of-the-art, open source voice model built for cinematic use cases. Traditional TTS gives you a voice. DramaBox by @resembleai gives you a performance. For too long, Voice AI has been stuck in "robotic assistant" mode. If you wanted dramatic emotion, sighs, or a voice cracking with grief, you had to hire an actor or spend hours editing. We fixed that.

introducing Parafield, the spatial sound intelligence. the FIRST agentic platform that combines voice, music, and sound effect, to create a personalized 3D sound experience for you. if you're interested in sci-fi, arts, and/or spirituality experience, this is for you. comment "sound" to access early beta. watch it with 🎧

We've published a paper that explains our views on AI competition between the US and China. The US and democratic allies hold the lead in frontier AI today. Read more on what it’ll take to keep that lead: https://t.co/TgJBeodWYK

More info at https://t.co/enHVCyIz53

Introducing Vesta The AI for your closet Vesta gives you daily outfits based on your wardrobe, your schedule, the weather, and your style. The days of having “nothing to wear” are over. https://t.co/x8IJiUv78V

NEWS: OpenAI hit with Class-Action Privacy Lawsuit for Sharing ChatGPT Data with Google and Meta. Sam Altman's OpenAI secretly embedded Meta’s Facebook Pixel and Google Analytics into ChatGPT, turning your most private conversations about health, finances, legal issues, and confidential company data into ad-targeting data sent straight to Meta and Google without consent. This violates federal wiretap laws. OpenAI enabled surveillance for profit.

Sanders and AOC introduced a bill to pause ALL AI data center construction. 300+ local bills filed. Half of planned 2026 data centers facing delays or cancellation. Each one brings billions to local economies. The people who say they want American jobs are trying to block the biggest job creation engine since the interstate highway system.

Check it - can fill in erased bits and denoise partially noised images! https://t.co/1ujiBMsK8v

Some repair snapshots https://t.co/6zs685DcWT

+1 more

Dream Air SE combines lightweight design, micro-OLED visuals, and eye tracking into an accessible next-gen VR headset starting at $899. Pre-order now to receive exclusive pre-order benefits: https://t.co/iv4ySczJWi https://t.co/GJd6fHLLW3

Dang I love this new world of hand-waving software into existence https://t.co/JOTYY1LUwJ

OpenClaw users who bought a Mac Mini right now: https://t.co/CYTSVPOfY3

me 10 mins into looking at agentic/harness benchmarks. https://t.co/CLkItyCwLy

me 10 mins into looking at agentic/harness benchmarks. https://t.co/CLkItyCwLy

OK may as well try the obvious one: can NCA learn to denoise? Can they generate images? We'll find out :) https://t.co/WNA2FdnJUf

I'm also training them to infill-noise, for the sole reason that it'll make a killer demo later :) https://t.co/bfWL0QVZAT

Yesterday we committed a cardinal sin: two first-party events in NYC, back-to-back. We had to close registration for both early. Packed rooms. Strong vibes. No regrets. 💻 Laptops out — developer workshop led by Jerry Liu and Logan Markewich 🍷 Glasses up — AI engineer happy hour with our friends at Tabs 🤝 Thanks to everyone who showed up. NYC 🗽 , you delivered.

Looks like @NVIDIAAI has released official getting started guide for hermes on DGX Spark! Great way for new folks to get started and believe in local agentic AI https://t.co/rbH8VJVa3p

@aaryan_kakad @jun_song They are the most complete of tech industry here on X and build this: https://t.co/kiuZ7QXLzb Love!

It’s happening… https://t.co/oBWoF1Wuye

It’s happening… https://t.co/oBWoF1Wuye

I mean just look at this beauty! seriously well deserved finalist @evvaaannnn in the @NousResearch x @Kimi_Moonshot creative hackathon! https://t.co/TVCuSkD1xK

YouArt (@YouArtStudio) empowers storytellers to create AI films and series with leading video agents, get funded by early fans, and earn subscription revenue. Congrats on the launch, @zhenghy723 & @edmakethings! https://t.co/2OEWWS4cb2 https://t.co/0fns35epYX

New course: Transformers in Practice. You'll get a practical view of how transformer-based LLMs work, so you can reason about their behavior, diagnose problems like slow inference, and make smarter decisions about deployment. This course is built in partnership with @AMD and taught by @realSharonZhou. You'll see how transformers generate text one token at a time, how the model decides which earlier words matter most when predicting the next one, and how techniques like quantization speed up inference on GPUs. This is not a video-only course; interactive visualizations throughout let you play with these concepts and build intuition that sticks. Skills you'll gain: - Understand why LLMs hallucinate, and RAG and chain-of-thought shape what they generate - Look inside the model to see how attention and layers combine to predict the next token - Diagnose inference bottlenecks and learn the techniques that speed up transformers on GPUs Join and understand what's really happening inside your LLMs: https://t.co/oS6ekeHsIw

Oof. One of the few things Americans of all parties appear to agree on. https://t.co/kMxKezGBan

Oof. One of the few things Americans of all parties appear to agree on. https://t.co/kMxKezGBan

Two can't-miss events coming up in Seoul: 👉 Modular's Judy Heflin presents an intro to Mojo & MAX at the Efficient AI Offline Meetup on May 16th at Dream Plus Gangnam: https://t.co/w9lpuL3EJq 👉 Our inaugural Modular Developer Meetup in Seoul on May 19th at Belgium Jazz Cafe: https://t.co/5ztBwpJeFv Come for the talks from Modular & SqueezeBits. Stay for the NVIDIA RTX 5080 + AMD Radeon RX 9070 raffles 👀

Today we’re releasing Toto 2.0: a family of open-weights time series foundation models spanning 4M to 2.5B parameters. The question we set out to answer was simple (yet previously open): Do time series foundation models get reliably better as they scale? Our answer: yes! 🧵 https://t.co/vo4zeAf7wD

More than 140,000 fake citations across four research repositories were identified in papers and preprints published in 2025 alone https://t.co/nEdDM4JBVD

【速報】 Hermes Agent に "Computer Use" が来た🤯 https://t.co/ZPWWH1HwH4 これ何が変わったか👇 ・どのモデルからでも実 PC を操作できるように 　→ Claude / GPT-5.5 のような特別モードに縛られない ・完全バックグラウンドで動く 　→ マウス / キーボードを奪われない ・ユーザーは普通に作業を続けられる 　→ "AIが裏で雑用、自分は本業"の構図が成立 つまり 「Computer Use = フロンティアモデル限定」 だった常識がここで終わる。 ローカル Qwen / Llama でも Computer Use が使える。 これ、Anthropic の Computer Use を待つ必要がなくなる レベルの破壊的アプデ。 OpenClaw でメモリ運用してる人は 明日からこの "裏 Computer Use" 連携を組み込み始めて損なし。