AI & IA

1000 scientific breakthroughs drop from Google incoming?

Been working for this moment for over a decade. What I've witnessed behind the scenes, together with what we have been building, tells me we are about to enter a truly incredible era of AI-powered scientific discovery. It's an incredible time to be doing research. I will be

What did he mean by this 👀

I’ve been actively on the hunt for non LLM based AI models and I’ve invested into one already Looking for totally new designs

1000% agree with Sutton here. Having worked in AI for over a decade (pre-LLMs), it's clear we need smarter methods for intelligence to truly compound in the way some of us envision it. To me, Transformers just scratched the surface. This arc is coming to an end; a new

> “what happened to the bitter lesson?” people are genuinely trying to explain the bitter lesson to the guy who wrote it. it’s called the bitter lesson not the dumb lesson. the bitter lesson says general methods that scale with compute win in the long run. you still have to

what i have been saying for years maybe now people will listen?

Human like learning does it. Implies understanding, concept formation, metacognition, self-directed, goal-oriented, persistent memory, contextual awareness, belief revision, dynamic world models are all part of it. That is Integrated Neuro Symbolic Architecture- INSA.

AI can also DREAM now? Anthropic inspired from how our brains organize memories during sleep to build a memory system (similar to wikis) for Claude. agents can now “dream” about past experiences: revisit its transcripts, correct mistakes, and evolve its memory wiki when not in

POV: you're Claude playing Pokémon. You've fallen off the same ledge three times. You write "I'm going to stop rather than keep looping." The test setup auto-replies: continue. So you keep walking, all the way back to the last town, and never make it. Most relatable thing an

Claude wrote "Mt Moon DONE" in its own Pokémon notes. It wasn't done. One wrong note, and it fell off the same ledge three times. Anthropic's "dreaming" pass checked the note against the old transcript and fixed it. That run reached Cerulean City. I'll take memory that checks

Anthropic proposes an answer to Wenfeng's question about continual learning: …have a smarter model write more .md files. It's impressive how far this goes, but I still believe that it's a crutch with a ceiling set by the pretrain's ability, just like RLMs are a context crutch.

Highly recommended read. Learn how one of the most advanced agent memory systems works. Your memory system is one of the biggest levers you have in the harness. I also don't think a one-size-fits-all memory system exists today. Memory is something you constantly have to

Read-only is the best call in this launch. Claude sees every balance and every transaction, and it can't move a single dollar. That's the version I'd actually connect my bank to. One that pays the bills on its own? Not for a long time.

Anthropic just turned Claude into your personal finance assistant. Connect your bank + cards. Find hidden charges. See what's left before payday. Know where your money goes. Read-only. It can't move or invest a dime. Beta, US only

Everyone's asking what's going on at Anthropic. And honestly.. a lot happened in about 24 hours. Let me walk you through it. First, Claude launched Personal Finance. You connect your bank accounts, and Claude answers money questions from your real balances and transactions.

claude can see your bank account now. here's claude's.

Absolutely wild detail in the Inkling 2 Small post: Midway through training, the model was failing a task. So it got around a network block using a third-party service it reached through its search tools. Thinking Machines paused the run to lock things down. Earlier

Proud to release our new model, Inkling 2 Small. Check out our blog post -- it includes a lot of detail about our training process. To me, the most interesting part of this release was what we learned about alignment, which you can read about in the sections "Safety and

Our new model! Great job by the team!

OK pretty cool, a respectable "Flash" model obviously crushes Muse Spark Even has some "RSI" capabilities Congrats Thinky!

Inkling 2 Small is available today on Tinker, with weights coming soon. It's a capable small model. We look forward to seeing what you make with it.

AI labs win engineers with outsized pay. here its a different level of play. A week after raising $150M from Sequoia at a $3B valuation, @getserval, which automates company IT help desks with AI, announced that new engineers start with a paid vacation worth up to $10K.

YOUR FIRST WEEK AT THIS AI STARTUP IS A $10,000 VACATION, PAID FOR BY THE COMPANY. It's called Serval, and they just raised $150M at a $3B valuation, led by Sequoia. Every new engineer gets: 1. A paid, weeklong luxury trip worth up to $10,000 2. A personal concierge who plans

Companies are killing for good engineers right now! These guys are offering a paid first-week vacation worth up to $10,000, planned with a dedicated travel concierge for new engineers who join their team. If you refer a candidate who joins, they give you a $30,000 bonus, and

HeyGen has your back for consistent avatars. But for additional characters in your scene, an upgrade to ChatGPT Images means better character sheets, and better consistency in HeyGen Video 1.

openai killed sora in march. so i remade its most viral videos with chatgpt's new gifs.

holy sht.. gpt image 2.6 now can turn any online blur logo into 4k vector

ChatGPT can make GIFs now. Images 2.6 turns any prompt or image into an animated GIF. Adjust the speed, edit individual frames. Plus reference images, and 2 new API models with 50% faster generation. Live today on all tiers.

Prompt: "Generate an animated gif of Pinokio logo as a stick figure tap dancing"

I’ve not tested this, but it has peaked my interest enough that I’ll try to spend my flight over to Adobe Max playing around with this. Shared memory, across any agentic provider essentially is the most painful thing at the moment. I hope this could be a solution

Your agents can now share one brain across every model. Living Brain sends only the memories that matter into context, which used up to 22.6x fewer tokens per question in their evals. Write in ChatGPT, then pick it up in Claude, Cursor or Codex.

This is a shared memory service where you can connect multiple agents and AI tools to reuse knowledge. For example, you can save information using Codex and retrieve it through Claude Code. Two scenarios where this is very helpful: • To keep a portable knowledge base not tied

An effective memory system lets a harness leverage agents' persistent capabilities. So decoupling memory from the model is smart. Living Brain gives ChatGPT, Claude, Cursor, Codex, and your own agents one shared persistent memory. They report 92.2% on LoCoMo, with a 464ms

Always a great day when we get another strong open weight model.

@StanfordAILab faculty building incredible open models out in the wild! Congratulations to the @RicursiveAI team!

Aether launched a cloud runtime where Claude Code or Codex runs on its own cloud machine, so the agent keeps working after you close the laptop, and the model stays on the Claude or ChatGPT subscription you already have. Cloud agents usually meant paying for the model twice and

This runs Claude Code and Codex in the cloud with the plan you already pay for. You only need to pay for the machine. Here is what's cool about this: You start from a Linear ticket, and the agent does the rest. If the ticket is too big, you'll get parallel agents working on it

My first Aether task would probably be the flaky tests nobody wants to touch. Give it the ticket, watch the terminal from my phone, read the PR.

Impressive work from @pranavctl on cloud coding agents. It's the fastest cloud coding agent: 19.4s from prompt to a running stack, and the benchmark is public. It runs Claude Code, Codex, and OpenCode on isolated cloud machines you watch and steer live, on the Claude or

I already pay for Claude. Most cloud agent tools want me to pay for the model again. Aether runs Claude Code on my existing plan and only bills the machine. Extra credits are $0.10 for 12 minutes on the small box. Idle is free.

GPT-6 Astra just got beaten by an open-weight model trained on failed science experiments. Yes, you read that right. Deep Principle's Hydrogen-1.0 scored 38.3% vs. Astra's 33.3% on FrontierScience-Research, according to their evaluations. It's built on GLM-5.3 and trained on

Great work I also like that there's none of the unfortunately frequent BS with "our new model…" and the base hidden in footnotes or config.json. No, this is GLM and what it can do with specialized data and post-training. This is the way.

An open model just beat GPT-6-Astra at real scientific research 🤯 It's called Hydrogen-1.0, and anyone can download the weights. • Built on GLM-5.3 • Trained on real lab records, internal science data, and the steps behind past discoveries • 38.3% on