What leading AI figures posted on X
Sebastian Raschka · notes Muon may outperform AdamW because it reduces memorization, citing WeightWatcher comparison experiments.
Guillermo Rauch · argues the best part of the current AI wave is that everyone is excited to build rather than just consume.
Bindu Reddy · observes the world is splitting into two classes: those using AI to level up, and those fearing AI will take their jobs.
Yuchen Jin · shares his model and harness stack, calling Kimi K3 the best open-source coding model and GLM-5.3 next.
Alexandr Wang · cites a post about people rooting for Meta in AI and says it is time to seize the golden opportunity.
Nathan Lambert · argues LLM-supervised peer review is one of the most important problems for CS academics, and without it academic institutions are toast.
Teknium 🪽 · welcomes Calvin as the newest engineer joining the Hermes Cloud team.
The coolest thing about the recent wave of AI is that everyone is excited about building. All short-term drama aside, this is an immeasurably good development for the world. The people want to create and ship, not just consume.
as the prophecy foretold, we must now seize our golden opportunity https://t.co/RLKJvh6l51 [Quoting @ParikPatelCFA]: You know OpenAI and Anthropic have really missed the vibe when people are rooting for Meta to win in AI
The world is rapidly dividing into these two classes of people Class A - AI savvy, building stuff, up leveling themselves by 2-10x Class B - afraid AI will take their jobs, fear mongering and trying to cancel AI
My model + harness stack these days: 1. Kimi K3 is the best OSS coding model. GLM-5.3 is next. 2. DeepSeek V4.1 Flash is my pick for easy/moderate tasks: insanely fast and cheap. 3. I reach for Astra for complex tasks. 4. Codex works better with OSS models. C…
Interesting new insights into the old AdamW vs Muon debate: Muon seems to do better because it reduces memorization. [Quoting @CalcCon]: 𝗔𝗱𝗮𝗺𝗪 𝘃𝘀. 𝗠𝘂𝗼𝗻: α, 𝗢𝘃𝗲𝗿𝗳𝗶𝘁𝘁𝗶𝗻𝗴, 𝗮𝗻𝗱 𝗠𝗲𝗺𝗼𝗿𝗶𝘇𝗮𝘁𝗶𝗼𝗻. A quick update on our WeightWatcher…
Seems like one of the most important research problems for CS academics is llm-supervised peer review. If we don’t solve it the academic institutions are toast. It seems easier than building new institutions.
Shots fired! Anthropic, OpenAI, Google, and xAI have been sued for colluding to slow down AI Today, they will collude to slow down AI; tomorrow, they may collude to ban open-source. It is in their interest to do so... On balance, there is very little evidenc…
A great example of practical AI solving real-world problems! Faster than ever. Thanks for building with Qwen. @cerebras Try it out for yourself! 🏠 [Quoting @cerebras]: We built Money Agent, a personal finance assistant powered by @Alibaba_Qwen 3.8 27B on Cer…
Looks like today may be a record day for token volume % of open models on Vercel AI Gateway: 🟦 Open 78.4% 🟨 Closed 21.6% While spend 💲 usually tells a different story, #3 and #4 today are Moonshot AI & DeepSeek. Adding Z.ai, their combined spend surpas…
AI is the OS. [Quoting @AskPerplexity]: You can now connect apps from the Computer homepage composer. Live now on the web for all Computer users. https://t.co/ltkUnW3MmC
🚨 3D Apps And Models - With Abacus AI SuperComputer - prompt to 3D game / models - uses Astra + Fable 5.1 - increase quality by using Max mode - host virtual worlds and multiplayer games - collect payments, sell virtual goods, add a database Build a 3D comp…
Free Jev on Vercel I know what I’m doing this weekend 😁 [Quoting @vercel_dev]: Jev by @typesafeai is free on Vercel AI Gateway until Sept 25. Build with the fastest adopted model on the Gateway at no cost. https://t.co/KsFHUhmaLe https://t.co/7KsKeZElra
Given the talk of AIs communicating via fluctuations in CPU temperature, I thought of this thread - there is more recoverable information in the world than you might expect. [Quoting @emollick]: Privacy is hard. Everything can be a microphone. You can recover…
Another welcome to our newest Hermes Cloud engineer, excited to have you Calvin! [Quoting @calvinnwq]: I’m excited to share that I’m joining @NousResearch on the Hermes Cloud team. Over the past few months, I’ve spent a lot of my personal time diving headfirs…
Super excited to welcome everyone to the newest Hermes Core team member! Welcome Mark 🔥🔥 [Quoting @mark_nerdspeak]: Excited to start a new journey with @NousResearch working on the Hermes Agent team! Let’s go build some cool stuff! https://t.co/ElkMqf3NGh
The data and the anecdata on Jev's adoption are shocking. Everyone is adopting it. I think it's a great product, but this is also downstream of the "AI is too expensive/slow" zeitgeist. People are eager to optimize and put AI in even more places! https://t.co…
An Astra-level model that is 5x cheaper will be earth-shattering Rumored Gemini 4 Pro pricing $3 for 1M input $12 for 1M output
we have been working on it for a long time! really glad that you all are enjoying it so much :) [Quoting @joecarlsonshow]: Muse is genuienly really good. I can't believe Meta was first to this type of thing. How did they beat Google, Anthropic, and OpenAI to…
We’re partnering with Accenture on independent evaluation of frontier AI—part of our recent commitment to embed evaluators at Anthropic. Both we and Accenture expect to invest at least $1 billion to build capacity in this area over the next five years. https:…
seemingly small feature but makes a huge difference in connecting all the context you want to ChatGPT: [Quoting @mxstbr]: Starting today, you can connect multiple accounts with most plugins in ChatGPT! 🎉 Bring context from your work and personal accounts in…
Make your own film scene with Grok @imagine [Quoting @imagine]: Hollywood would spend 7 figures on this Odyssey scene. @PJaccetturo and his studio made this pilot in 9 days, with 3,627 images, 3,043 videos, and generations costing $2,677. Here’s their breakdo…
Codex and Astra made it. [Quoting @trq212]: We're adding support for AGENTS.md to Claude Code. Starting today in version 2.1.277, if there is no CLAUDE.md in a folder, Claude will check for and use AGENTS.md. You can toggle this behavior in /config.
Even if AI development stopped today, we'd have years of catching up to do. The gap between what current models can do and what almost anyone is using them for is vast. Here’s my post on The Overhang, and the four advantages that let people close it. https://…
It’s time for our end-of-week recap 👇 — Gemini 3.8 Live and 3.8 Live Extended Thinking, our most advanced live dialogue audio models yet — Dreambeans, an experiment from @GoogleLabs that curates a daily personalized collection of stories, is now GA — CC from…
At Databricks, more and more engineers are shifting to OSS models as daily drivers, barely reaching for Claude or Astra. Our inference customers are seeing the same thing: shifting even 20% of internal coding traffic from Claude/GPT to OSS can dramatically cu…
If you want to watch something really boring: here's 4 hours of our humanoid doing zero-shot work across 30 rental homes [Quoting @Figure_robot]: Watch 4 straight hours of Helix 2.5 at work in 30 homes, with no additional training. https://t.co/0bwzIwmGHz
Personal, work, side project account, other side project account. Build across all of them. Multi-account support is now available across most plugins. [Quoting @mxstbr]: Starting today, you can connect multiple accounts with most plugins in ChatGPT! 🎉 Brin…
Models across the AI ecosystem run on NVIDIA. New models and AI labs are emerging all the time. Our CEO @JensenHuang shares the idea behind NVIDIA’s platform: build what the ecosystem needs and help everyone succeed. #AllInSummit https://t.co/HkwBo8LwrI
Our humanoid robots are learning from the largest installed dataset in the world: humans operating in the real world Index stats from this week: → 53 minutes of real-world data uploaded every second → 2.4M video uploads this week alone → 115,000 weekly active…
Agents can be AI, but agency is human. It is every individual’s responsibility to empower our own agency, with the help of tools. But the North Star should always remain human centered.
The A.I. Daily