Quantum computing has long promised to revolutionize how we solve some of the most complex problems in science and technology, yet actually proving its practical edge over classical supercomputers has remained elusive. Recently, I came across fascinating insights about a milestone that finally brings us closer to that promise. The Quantum Echoes algorithm, running on Google‘s Willow quantum chip, has accomplished the first-ever verifiable quantum advantage on actual hardware — not just simulations or theoretical proposals, but real devices.

This breakthrough means that a quantum computer has now done something an incredibly powerful classical supercomputer just can’t match, and it’s repeatable and checkable. The implications for fields like medicine, materials science, and chemistry could be huge.

What is quantum advantage, and why does verifiability matter?

Quantum advantage refers to a quantum computer solving a problem faster or more efficiently than any classical computer could. Back in 2019, the first evidence of such advantage was demonstrated, but those early successes were limited to contrived problems without clear real-world impact and lacked built-in verification. Fast forward to today, where the Quantum Echoes algorithm not only runs 13,000 times faster than top classical supercomputers but also produces verifiable results—meaning the output can be reliably reproduced and checked on other quantum systems of similar scale.

This is huge because reproducibility is the cornerstone of practical usefulness. If quantum results can’t be verified, their reliability and utility remain questionable. Now, with the ability to cross-check computations, we can trust that quantum computers have truly outpaced classical machines in meaningful ways, a vital step towards trustworthy real-world applications.

How Quantum Echoes works: the power of a quantum “echo”

The Quantum Echoes algorithm essentially uses a clever trick: it sends a quantum signal forward through a system of qubits, tweaks one qubit along the way, then runs the operations backward to catch an “echo” of that disturbance. This echo is amplified through constructive interference, making it exquisitely sensitive to how information and disturbances spread across the quantum chip.

This four-step cycle—running operations forward, perturbing a qubit, running them backward, and measuring—is a sophisticated way to probe quantum dynamics with precision. Thanks to Willow’s 105-qubit array with ultra-low error rates and high-speed gates, this algorithm goes beyond previous quantum benchmarks focused just on complexity and steps toward precision calculations. It’s a new class of quantum challenge that mimics real physical experiments.

From quantum advantage to real-world applications

Quantum computing isn’t just about speed; it’s about unlocking new ways to understand nature, especially at the atomic and molecular level. Nuclear Magnetic Resonance (NMR) spectroscopy is a key tool in chemistry and biology for revealing molecular structures, but it has limitations in sensitivity and range. The Quantum Echoes approach acts like a “molecular ruler” that can measure longer distances and extract richer information from NMR data.

In partnership with the University of California, Berkeley, this technique was tested on molecules with up to 28 atoms. Astonishingly, the quantum results matched traditional NMR while also providing new structural insights unavailable by conventional means. This hints at the potential for a “quantum-scope” capable of unveiling details of molecular and material structures that were previously hidden from us.

Such advances could dramatically impact drug discovery, where understanding how a drug binds to its target molecule is crucial, or materials science, for designing better batteries, polymers, or even new quantum hardware components themselves.

Key takeaways to keep in mind

• Verifiable quantum advantage means that quantum results are reliably repeatable and checkable, a critical step for practical applications.
• The Quantum Echoes algorithm leverages a quantum echo phenomenon to measure the spread of disturbances with unprecedented precision on a 105-qubit chip.
• Applying quantum computing to enhance NMR spectroscopy opens a path toward new molecular and material insights with real-world implications for science and industry.

Looking forward, this breakthrough is more than just a technical feat. It signals the dawn of a new era where quantum computers start delivering tangible benefits beyond academic milestones. As quantum hardware improves toward long-lived, error-corrected qubits, we can expect more powerful algorithms that open new frontiers in medicine, chemistry, and materials science.

It’s exciting to witness how the once futuristic vision of quantum computing is steadily transforming into a practical tool that could redefine how we explore and harness the natural world.

The post Google Quantum Echoes: The first real proof that quantum computing actually works appeared first on Aiholics: Your Source for AI News and Trends.

What if there was a virtual version of you — a digital twin of your brain that doctors could use to test treatments before you even try them? Sounds like science fiction, right? Well, even as a neuroscientist, I find this notion mind-blowing. So today, let’s deep dive into this fascinating idea, explore how digital twins are transforming brain care, and glimpse what this could mean for the future.

First off, have you ever wondered how engineers predict whether a rocket will launch successfully or how factories optimize every machine on the floor without physically testing every scenario? They use something called digital twins. Simply put, a digital twin is a virtual replica of a real-world system that mirrors its behavior by constantly receiving data and using advanced models to simulate outcomes.

Imagine you’re designing a new vacuum cleaner. Instead of building dozens of physical prototypes, you can create a digital twin and run hundreds of simulations on how it performs under different conditions. This speeds up development, saves costs, and highlights potential failures well in advance.

Digital twins, originally pioneered by NASA for Apollo missions, are now everywhere—from simulating jet engine stress in aviation to optimizing robots on automotive assembly lines. Even smart cities use them to predict traffic jams and manage energy consumption.

So naturally, healthcare would be a perfect place for digital twins to flourish. And in fact, they already are. In cardiology, virtual heart models combine wearable ECGs and imaging data to simulate different pacemaker settings so doctors can personalize treatments without guesswork. During the COVID-19 crisis, lung models helped ICU teams predict pneumonia progression and optimize ventilation strategies. Orthopedic surgeons build digital replicas of knees or spines to experiment with prosthetics before surgery, improving fit and recovery times.

All these examples are exciting, but they involve relatively simpler organs. Now, what about the brain—the most complex organ in our body? Could we build a reliable digital version?

Modeling the brain is like simulating all the traffic in New York City—from taxis to pedestrians—in real time. It’s a chaotic, massive challenge because the brain has roughly 86 billion neurons firing electrical and chemical signals nonstop. To truly mimic it, we need data that spans genetics, molecular pathways, brain wiring, electrical activity, blood flow, plus behavior and environment.

The promise of brain digital twins is huge. Imagine testing treatments for epilepsy, predicting Alzheimer’s years before symptoms, or rehearsing delicate neurosurgeries on your personal virtual brain rather than risking the real one.

Before AI became mainstream, early computational neuroscience gave us mathematical models of neurons and brain regions, as well as brain atlases mapping structures at population levels. But these didn’t capture individuals or update in real time. Back in 2013, a team using the world’s most powerful supercomputer managed to simulate just a second of 1% of brain computation—and that took 40 minutes!

Fast forward to the AI era, and things are moving fast. For example, combining MRI structural data with EEG recordings, researchers now build personalized brain network models that simulate seizures, helping surgeons identify epileptic zones with unprecedented precision. In neurodegenerative diseases like Alzheimer’s and multiple sclerosis, AI models analyze longitudinal MRI scans to flag abnormal brain shrinkage five to six years before symptoms appear—opening a critical window for early intervention.

In surgery, AI-powered segmentation of MRI and CT scans creates interactive 3D brain twins complete with tumors and blood vessels. Surgeons can practice virtual procedures using VR headsets, receiving real-time feedback and suggestions to avoid critical areas. It’s like rehearsing the most complex operation before a single incision.

But it’s not all smooth sailing. Building and running brain digital twins is seriously challenging. First, acquiring the vast amounts of high-resolution, continuous data needed—think repeated MRIs, EEGs, behavioral inputs—is expensive and sometimes practically impossible. Then there’s the massive computational demand; simulating even a tiny network of neurons requires huge GPU clusters or supercomputers. Real-time, full-brain simulations remain out of reach today.

Beyond tech, the ethical and privacy concerns loom large. A brain twin holds your most intimate data—thought patterns, risk profiles, even personality markers. Ensuring patient consent, data ownership, and airtight security is critical. We also need to be vigilant about bias: if the data used to train AI systems is skewed toward certain populations, these models could unintentionally reinforce healthcare inequalities.

Despite these hurdles, the trajectory is clear. The future looks like hybrid digital twins, which combine detailed models of key subsystems (like memory or motor control) with higher-level abstractions for the rest. This strikes a balance between accuracy and scalability. We can also expect virtual clinical trials on cohorts of digital brains—speeding up drug and neuromodulation research, and lowering costs.

Other exciting prospects include real-time surgery overlays and digital mental health coaches that monitor mood and cognition through wearables and smartphones—providing early warnings and personalized interventions for conditions like depression and dementia.

The dream is a world where your doctor doesn’t just describe treatment options, but shows you exactly how each choice plays out on a virtual you—making medicine faster, safer, and truly tailored.

So does this all sound super sci-fi? Maybe, but it’s becoming a tangible reality faster than we might think. If brain digital twins intrigue you, there’s plenty more futuristic neuroscience to explore—like how brain chips are already making waves in clinics. The future of personalized brain care is just getting started, and it’s an incredible ride to watch.

The post What if AI could create a digital twin of your brain? Exploring the future of personalized brain care appeared first on Aiholics: Your Source for AI News and Trends.

Okay, AI fans, we’ve gotta talk about something pretty exciting that just dropped in 2025: Z.AI’s GLM 4.5 series. If you’ve been following open-source AI, you’ll know it’s rare to see a release this powerful, efficient, and accessible all at once. But that’s exactly what the folks at Z.AI (formerly Zepoo AI) have pulled off. From blazing-fast speeds and giant context windows to nuanced agent capabilities—all while being incredibly affordable—it’s shaping up to be a game changer.

Why GLM 4.5 is turning heads

Let’s start with the basics. GLM 4.5 is a huge foundation model with 355 billion parameters, but here’s the clever bit: it uses a mixture of experts architecture. That means not all parameters fire at once during inference. Instead, just 32 billion parameters are active per prompt. That design helps balance the heavy lifting with cost-efficiency and makes it possible to run powerful models without astronomical compute resources.

If you aren’t sitting on a supercomputer, no worries. Z.AI also released GLM 4.5 Air, a leaner sibling with 106 billion total parameters and 12 billion active, tailored for consumer-level GPUs with 32 to 64 GB of VRAM. So whether you’re a researcher, developer, or just an AI enthusiast with accessible hardware, Z.AI is throwing a bone here.

Built for autonomous agents and real-world use

GLM 4.5 is not just another chatbot. It’s engineered from the ground up as an autonomous agent with deep reasoning skills. It can:

• Think step-by-step over multiple turns
• Call APIs and interact with external tools
• Control interfaces and plan actions

The model offers two distinct modes—one optimized for deep, slow, complex reasoning, and another tuned for quick, speedy responses when you just want an answer fast. This hybrid approach baked into the architecture makes GLM 4.5 flexible enough to work across a wide range of practical applications.

And when it comes to speed, GLM 4.5 is seriously impressive. Thanks to speculative decoding and multi-token prediction layers, it can generate more than 100 tokens per second through its API—going up to 200 tokens/second in ideal scenarios. For context, the model supports a colossal 128,000-token input context window and 96,000-token output window, which dwarfs most competitors like GPT-4 or Claude 2.

The secret sauce behind training and architecture

Training a model this capable took some serious innovation. It started with 15 trillion tokens of general pre-training data, followed by an extra 7 to 8 trillion tokens focused on code, reasoning, and agent tasks. But Z.AI didn’t stop there—they rolled out a custom reinforcement learning system dubbed Slime, which optimizes both synchronous training and asynchronous rollout simulations, all while keeping GPUs efficiently utilized—even when dealing with slow, multi-step agent actions.

The architecture itself opts for depth over width—more layers with narrower hidden dimensions, favoring better reasoning capacity. They also threw in grouped query attention, partial rotary positional embeddings, and bumped to 96 attention heads for a hidden size of 5,120. It sounds complex, but this translates to better performance on demanding benchmarks without destabilizing training.

Benchmarking: Top tier but affordable

On major benchmarks, GLM 4.5 isn’t just competitive—it’s among the very best. It ranked third globally across 12 big tests involving reasoning, math, coding, and agentic behavior. Beating out models like Claude 4 Opus in many tests, and sitting just behind the giants GPT-4 and XAI’s Gro 4, it’s clear that Z.AI’s approach pays off.

For example, it scored an impressive 91% on AIM 24 reasoning and 98.2% on Math 500. Coding benchmarks show a 53.9% win rate over Kimmy K2 and an 80.8% success rate beating Quen 3 Coder. Plus, its tool calling success rate of 90.6% outperforms several peers by a noticeable margin—crucial for agents that need to work autonomously with external APIs.

And here’s something you’ll want to hear: the API pricing is incredibly low—roughly 39 cents per million tokens combined input/output in USD terms. That’s less than a tenth of the price of competitors like Claude, making high-level AI accessible at a price point that could truly broaden adoption.

Open source and user-friendly deployment

The best news? GLM 4.5 is fully open source under the MIT license. You can grab the model weights, run it locally, customize it, or integrate it into your own stacks. Its compatibility with existing AI agent frameworks and OpenAI-style APIs makes swapping or testing it painless—exactly what businesses and researchers want when experimenting with new tech.

Z.AI is also showcasing full demos that show off real power. We’re talking about AI that can research topics online, build and manipulate games like Flappy Bird, generate polished slide decks, and even create full-stack web applications on the fly with multi-turn conversational refinement. The code is clean, functional, and user-friendly—a huge leap from clunky AI prototypes we’re used to.

The bigger picture: China’s push in open-source AI

Z.AI’s move is part of a broader trend in China’s AI landscape, where startups like Moonshot, Step Aai, and Bichuan are racing to release cutting-edge open models, challenging the dominance of expensive, closed US models like GPT-4 and Claude 3.

With deep pockets from Tencent, Alibaba, and local governments, Z.AI isn’t just throwing a stone—they’re gearing up to lead with plans for an IPO and continued heavy investment in foundation models, multimodal capabilities, and more. Their fastest follow-ups are already underway, signaling a long-term bet on accessible, powerful AI for developers and businesses around the world.

Key takeaways for AIholics

• GLM 4.5 uniquely balances scale, speed, and cost, enabling real-world deployment of cutting-edge AI without breaking the bank.
• Its design for autonomous agents represents a genuine leap, supporting reasoning, API calls, and multiturn planning baked into the architecture.
• Open source and commercial friendly licensing makes it an irresistible option for startups, researchers, and enterprises wanting flexibility and control.

Wrapping up

What Z.AI has done with GLM 4.5 feels like a pivotal moment in AI democratization. Powerful models with huge context windows, blazing speeds, agent capabilities, and low costs—plus open source. It’s a combo that has the potential to reshape the AI ecosystem and challenge the closed, pricey giants.

Whether you’re building autonomous agents, complex code assistants, or exploring novel AI applications, GLM 4.5 deserves your attention. It’s exciting to watch the open-source world catch up and even surpass some of the big industry players.

So what do you think? Could open-source models like GLM 4.5 topple the current closed heavyweights? Drop your thoughts below—I’m curious to hear your take.

The post Z.AI’s GLM 4.5: a breakthrough in open-source AI that’s fast, efficient, and affordable appeared first on Aiholics: Your Source for AI News and Trends.

The tech world is showing concern over a major decision taken by the UK government. There is about £1.3bn ($1.7bn) worth of projects that have been cancelled which were meant to enhance UK’s capabilities in artificial intelligence (AI). This has raised concerns that Britain might lose its position as an AI technology thought leader.

Key points

There are two main projects being abandoned. The first one was known as Artificial Intelligence Research Resource which aimed at improving the computing capacity of the United Kingdom on AI and was priced at £500m. On the other hand, there was another bigger project which cost £800m that would have seen a supercomputer being built at Edinburgh University. An exascale computer which could perform 1 trillion operations per second.

These were important projects because they would have given to UK what it needed to create and run advanced AI systems. These types of systems require a lot of computational power and data to function properly. If this cancellation happens then, therefore, Britain may lag behind among other countries in terms of AI.

The motivation for this decision by the new labour government was because it wanted to save money. They claim that they found a “financial black hole” from the previous government and need cut in many areas though this could affect their technological industry significantly in future

For now, Britain has been doing well with regards to AI. It has attracted many investments and numerous organizations have adopted AI technology into their systems including some big firms such as Google or Amazon…… According to recent researches, UK is among leading nations globally when it comes to generative AI adoption rates. In fact, without these new builds, keeping pace with other countries making large investments in AI becomes more difficult.

Related articles

Norway invests in AI research with new national supercomputer

November 2, 2025

AI Steve: Meet the first Artificial Intelligence candidate running for UK parliament

November 3, 2025

Many are scared about what will happen next after this announcement; thus they want government officials who are responsible for all these things come up with new plans immediately so that UK does not drop back into a weak position regarding AI and any other significant tech industries.

The government still claims that they are dedicated to technology innovation for national development. They are developing a new plan known as AI Opportunities Action Plan. However, the cancellation of these big projects has made many people in the technology sector wonder what will be next on the UK’s AI agenda.

The post UK cuts major AI funding, raising concerns about tech leadership appeared first on Aiholics: Your Source for AI News and Trends.

Norway is set to boost its AI research capabilities with a new state-of-the-art supercomputer. The 225 million NOK contract, awarded to Hewlett-Packard Norway AS (HPE), marks a significant step forward for the country’s research and innovation in artificial intelligence.

Minister of Research and Higher Education Oddmund Hoel emphasized the importance of this investment for Norway’s knowledge independence and national security. The supercomputer will provide the computing power necessary to keep pace in the global AI race, reducing reliance on foreign actors.

The new system, a Cray Supercomputing EX model, will feature 304 advanced NVIDIA GH200 GPUs, making it the most powerful in Norway’s history. This high-performance computing (HPC) capability will support data-driven research across various fields, including medicine, climate science, and language processing.

Gunnar Bøe, Managing Director of Sigma2, the state-owned company responsible for the national supercomputers, stressed the critical nature of this investment for Norway’s research future. The system will be accessible to researchers nationwide, regardless of their institution or field of study.

Sustainability is a key focus of the project. The supercomputer will be installed at the Lefdal Mine Datacenter on Norway’s West Coast, utilizing cold water from the nearby fjord for cooling. Despite its increased power, the new system is expected to reduce energy consumption by over 30% compared to previous generations.

The supercomputer is scheduled for installation in spring and summer 2025. This investment represents a crucial step in Norway’s commitment to advancing its AI research capabilities and maintaining competitiveness in the international scientific community. As Professor Stephan Oepen of the University of Oslo noted, this expanded GPU capacity is essential for developing large Norwegian language models and keeping pace with the rapidly evolving field of AI.

The post Norway invests in AI research with new national supercomputer appeared first on Aiholics: Your Source for AI News and Trends.