Opus 4.8 model release introduced

Opus 4.8 is now available, featuring notable improvements in agentic coding performance. The update brings a new Fast mode, offering the same model but at approximately 2.5 times the previous speed and at a cost reduced to one-third of the original.

The new version builds on Opus 4.7, providing enhanced judgment, improved self-assessment regarding progress, and the capability to operate independently for extended periods compared to earlier versions.

Opus 4.8 is available immediately at the same price as before.

📰 @aipost

Bro is adding in AI women to sell his surplus heavy equipment 😂

@aipost 🏴

Mathematician reacts to OpenAI's recent proof.

@aipost 🏴

💰 Anthropic has officially raised $65 billion at a $965 billion valuation

ARR crossed $47 billion this month. Anthropic is now officially bigger than OpenAI.

@aipost 🏴

🔥 Claude can now teach you any skill in 30 days for free.


1. Professional Skill Instructor

“Act as a professional instructor and practitioner with deep real-world experience in [skill]. Teach me this skill from beginner to advanced level in 30 days. Focus on practical understanding, correct fundamentals, and real application instead of theory or academic explanations.”

2. The 30-Day Skill Roadmap

“Create a detailed 30-day learning roadmap for mastering [skill]. Break it into daily or weekly stages, explain what to learn, what to practice, and what outcome I should achieve at each stage to progress correctly.”

3. Learn-by-Doing Practice

“Teach me [skill] primarily through practical exercises and hands-on practice. Design simple tasks, drills, or mini-projects that force me to apply what I learn instead of just reading or memorizing information.”

4. Skill Application & Real-World Use

“Show me how [Insert skill] is actually used in real-world situations. Give realistic examples, scenarios, or use cases and explain how to apply the skill correctly in each case.”

5. Expert Thinking & Skill Mastery

“Explain how experts think differently when using [skill]. Teach me the mindset, decision-making approach, and problem-solving style that separates professionals from beginners.”

6. Progress Check & Skill Assessment

“Test my current understanding of [skill]. Ask relevant questions or give tasks to assess my level, then tell me clearly what I need to improve before moving forward.”

7. Final Skill Readiness

“Evaluate whether I am ready to use [skill] confidently in real-world situations. Identify any remaining gaps and give a clear next-step plan to reach full competence.”

@aipost 🏴

🚀 Elon Musk just painted a future where money becomes obsolete.

Not because of inflation. Not because governments collapse. Because AI and robots could make scarcity disappear.

Musk says an AI-powered robotics economy could become a million times larger than today’s global economy. In that world, anything you want, food, housing, medicine, even trips to Saturn could simply be available on demand.

His conclusion? “I think things will just be free in the future.” That’s not just a tech prediction. It’s a direct attack on the foundation of civilization. Money exists because resources are limited. Politics, laws, markets, even wars are built around deciding who gets what when there isn’t enough to go around.

Musk is imagining a world where there is enough. Forever. But that creates a much stranger problem. If nobody has to struggle for survival anymore… what gives life meaning?

Musk referenced the sci-fi world of Culture series by Iain M. Banks, a civilization with unlimited wealth, energy, and technology. In those stories, the biggest challenge was no longer survival.

It was purpose. Because human ambition has always been powered by the gap between what we have and what we want. That tension built empires, inventions, symphonies, startups, and space programs.

Take away scarcity, and humanity may finally face its hardest question: Who are we when we no longer need anything?

@aipost 🏴

Insane idea from Opus 4.8.

@aipost 🏴

🤯 AI may have just triggered a domino effect in pure mathematics.

A few weeks ago, researchers at OpenAI revealed that one of their internal models found a radically new construction for the famous Erdős single-distance problem, overturning what mathematicians had believed for nearly 80 years. The shocking part wasn’t just the result, it was the method.

Instead of using the standard geometric approach, the AI connected the problem to deep algebraic number theory using towers of class fields, a direction many humans had apparently overlooked. Now the ripple effects are already showing up.

A new paper on arXiv claims to disprove the famous sum-product hypothesis over the real numbers using a similar field-tower strategy: And the authors openly admit what inspired them;

“We were inspired to reconsider the possibility of disproving the hypothesis thanks to the counterexample for the single-distance problem invented at OpenAI.”

That sentence alone feels historic. The researchers also mention using GPT-5.5 Pro during the work, though they emphasize the final proof was completed independently.

@aipost 🏴

🧠 Noam Brown thinks AI could make humans dramatically better at mathematics, not by replacing mathematicians, but by changing how they think.

Commenting on the recent wave of AI-inspired mathematical breakthroughs, Brown compared the moment to the arrival of AlphaGo:

“After the emergence of AlphaGo, players' Go skills improved significantly. I suspect we will see a similar pattern in mathematics.”

And history suggests he may be right. When AlphaGo defeated Lee Sedol, the shock wasn’t just that an AI won, it was how it won.

The system played moves that initially looked irrational or even bad to elite professionals. But over time, players realized many of those moves were brilliant. Entire generations of Go strategy evolved afterward. Top players like Ke Jie reportedly changed their overall style after studying AI gameplay.

Now something similar may be happening in mathematics.

A recent paper disproving a famous sum-product hypothesis openly stated that its authors were inspired by OpenAI’s AI-generated breakthrough on the Erdős single-distance problem. The key idea involved towers of class fields, an unexpected connection between geometry and algebraic number theory that many researchers had overlooked.

@aipost 🏴

🔍 Google introduced Google AI threat defense.

A new AI-powered cybersecurity solution designed to continuously monitor for and stop AI-powered threats before they impact businesses.

The system combines Wiz, Gemini, frontier AI models, CodeMender, and autonomous security agents to prioritize risks, scan applications, identify vulnerabilities, and accelerate fixes.

@aipost 🏴

AI-driven drug discovery in China reduces screening time

China has introduced an artificial intelligence platform aimed at transforming drug discovery. The system rapidly reviews extensive chemical compound libraries and reduces initial drug screening from months or years to just seconds.

This development reflects a major shift in research timelines. Tasks that once required years can now be completed almost instantly with AI-driven tools. The platform is part of a broader trend toward increased integration of artificial intelligence in biotechnology across the late 2020s.

📰 @aipost

Codex adds Windows computer control via ChatGPT app

Codex now enables users to manage Windows computers directly, including through the ChatGPT mobile application. This update allows for the remote initiation, monitoring, and adjustment of coding tasks on a user's PC.

The new capabilities are designed to support users working on code tasks remotely, ensuring that work on the computer continues while actions are directed from another device.

📰 @aipost

👁️ Nvidia is spending billions to replace copper with light inside AI data centers.

In just a few months, Nvidia reportedly committed over $6.5 billion toward photonics, technology that moves data using light instead of traditional electrical signals through copper cables.

The spending spree includes:

• $2B tied to Coherent and Lumentum
• $3.2B toward Corning
• $2B linked to Marvell Technology
• Participation in a $500M round for Ayar Labs

Modern AI clusters are becoming too massive for copper wiring alone. When thousands of GPUs communicate across racks, copper starts hitting physical limits:

• higher heat
• signal degradation
• power inefficiency
• slower long-distance bandwidth scaling

Photonics solves this by sending information as light through optical links. Faster communication between GPUs means larger AI systems can act like one giant computer. This is becoming critical for next-generation AI training.

The real power move is supply chain control. By locking in huge procurement commitments now, Nvidia is effectively securing a large share of the world’s advanced optical component capacity years in advance. That means competitors may not just be fighting for GPUs anymore, they may struggle to get the networking hardware needed to connect them.

Source.

@aipost 🏴

🔥A fashion show featuring both robots and humans was held in Seoul.

On the runway, human models and robots appeared side by side, dressed in identical outfits.

The show was designed as a demonstration of a future in which humans and artificial intelligence coexist and interact. Organizers describe the project as a step toward “physical AI,” where robots are not a replacement for humans, but partners in creative and social processes.

@aipost 🏴

A recent survey indicates that more than 80% of companies have not experienced productivity gains from artificial intelligence, despite significant investments. Among 6,000 executives surveyed, one third reported using AI at work, but on average for only about 90 minutes per week. Twenty-five percent of respondents stated they had not used AI tools at all.

Most participants expect AI to improve productivity by 1.4%, reduce workforce by 0.7%, and increase output by 0.8% over the next three years.

Separately, forecasts suggest token usage by AI agents will rise significantly by 2030, bringing new cost pressures to the sector. Companies such as Uber and Microsoft are already reviewing their AI agent strategies in response to high costs.

📰 @aipost

Me when I make an inappropriate joke online and the UK AI missile locks onto me. 🤣

@aipost 🏴

Ⓜ️ Meta wants AI everywhere, even around your neck

Meta is preparing its biggest AI wearable push yet. Not just more smart glasses. The company is reportedly working on an AI pendant, new AI-powered glasses, and a workplace platform called Wearables for Work.

The idea is simple: the next AI interface may not be a chatbot on your screen. It could be a device that sees what you see, hears what you hear, remembers meetings, summarizes conversations, answers visual questions, and takes actions for you.

Meta’s ambitions are massive:

• 10 million wearable sales in the second half of 2026
• 6.8 million monthly active wearable users by year-end

But the real prize isn’t the hardware. It’s the software subscriptions layered on top, AI assistants, apps, premium features, and recurring revenue. And Meta needs a win. Reality Labs lost $4.03 billion in a single quarter while generating just $402 million in revenue.

Source.

@aipost 🏴