STEM

[Cadavre Exquis]

Anthropic Releases Claude 4 Models That Can Autonomously Work For Nearly a Full Corporate Workday
Posted by msmash on Thursday May 22, 2025 @01:20PM from the moving-forward dept.

Anthropic launched Claude Opus 4 and Claude Sonnet 4 today, positioning Opus 4 as the world's leading coding model with 72.5% performance on SWE-bench and 43.2% on Terminal-bench. Both models feature hybrid architecture supporting near-instant responses and extended thinking modes for complex reasoning tasks.

The models introduce parallel tool execution and memory capabilities that allow Claude to extract and save key facts when given local file access. Claude Code, previously in research preview, is now generally available with new VS Code and JetBrains integrations that display edits directly in developers' files. GitHub integration enables Claude to respond to pull request feedback and fix CI errors through a new beta SDK.

Pricing remains consistent with previous generations at $15/$75 per million tokens for Opus 4 and $3/$15 for Sonnet 4. Both models are available through Claude's web interface, the Anthropic API, Amazon Bedrock, and Google Cloud's Vertex AI. Extended thinking capabilities are included in Pro, Max, Team, and Enterprise plans, with Sonnet 4 also available to free users.

The startup, which counts Amazon and Google among its investors, said Claude Opus 4 could autonomously work for nearly a full corporate workday -- seven hours. CNBC adds:

"I do a lot of writing with Claude, and I think prior to Opus 4 and Sonnet 4, I was mostly using the models as a thinking partner, but still doing most of the writing myself," Mike Krieger, Anthropic's chief product officer, said in an interview. "And they've crossed this threshold where now most of my writing is actually ... Opus mostly, and it now is unrecognizable from my writing."

Krieger added, "I love that we're kind of pushing the frontier on two sides. Like one is the coding piece and agentic behavior overall, and that's powering a lot of these coding startups. ... But then also, we're pushing the frontier on how these models can actually learn from and then be a really useful writing partner, too."

Saludos.

[Cadavre Exquis]

New sodium-based solid-state batteries may unlock faster charging and higher capacity
A novel NASICON electrolyte developed by BAM researchers may hold the key to stable, high-performance sodium solid-state batteries. Promising room-temperature operation, better safety, and lower costs.

Nathan Ali · 2025.05.20

BAM’s NASICON breakthrough pushes sodium solid-state batteries closer to real-world use (Image source: BAM)

Researchers at Germany’s Federal Institute for Materials Research and Testing (BAM) are re-engineering solid-state batteries to overcome the ceiling reached by today’s lithium-ion cells. Their project centers on a new sodium super-ionic conductor (NASICON) electrolyte that could unlock faster charging, longer service life and lower costs without compromising safety.

Conventional lithium-ion packs rely on graphite anodes that store a finite number of ions. Switching to metallic lithium —or the cheaper, more abundant sodium— would raise energy density by as much as 40 percent. The catch is that solid anodes need a solid electrolyte, and the rigid interface between the two often forms voids that disable the battery. A partially liquid anode can solve that interface problem, but only if the whole system stays stable.

BAM’s team, led by guest researcher Gustav Graeber, has already shown that a liquid alkali-metal anode can deliver 100 times the power of graphite. Right now, though, that record output appears only at about 250°C (482°F). To bring the technology down to room temperature, the researchers add potassium to lower the anode’s melting point. Most solid electrolytes degrade in contact with potassium, so the electrolyte becomes the new bottleneck.

NASICON materials break that impasse. They conduct ions well at ambient conditions and tolerate potassium, especially when doped with hafnium. Hafnium, however, is scarce and expensive. The BAM project therefore, screens earth-abundant dopants that can match hafnium’s stabilizing effect. The most promising compositions are already being integrated and cycled in prototype sodium cells.

If the search succeeds, sodium-based solid-state batteries could move from the lab to everyday devices and electric vehicles. Higher energy density would extend operating time, while solid electrolytes would improve intrinsic safety. Faster charging and a supply chain that leans on plentiful sodium rather than scarce lithium and cobalt would make the technology attractive for grid storage as well—an incremental but meaningful step toward lower-carbon energy systems.

Saludos.

[el malo]

New sodium-based solid-state batteries may unlock faster charging and higher capacity
A novel NASICON electrolyte developed by BAM researchers may hold the key to stable, high-performance sodium solid-state batteries. Promising room-temperature operation, better safety, and lower costs.

Nathan Ali · 2025.05.20

BAM’s NASICON breakthrough pushes sodium solid-state batteries closer to real-world use (Image source: BAM)

Researchers at Germany’s Federal Institute for Materials Research and Testing (BAM) are re-engineering solid-state batteries to overcome the ceiling reached by today’s lithium-ion cells. Their project centers on a new sodium super-ionic conductor (NASICON) electrolyte that could unlock faster charging, longer service life and lower costs without compromising safety.

Conventional lithium-ion packs rely on graphite anodes that store a finite number of ions. Switching to metallic lithium —or the cheaper, more abundant sodium— would raise energy density by as much as 40 percent. The catch is that solid anodes need a solid electrolyte, and the rigid interface between the two often forms voids that disable the battery. A partially liquid anode can solve that interface problem, but only if the whole system stays stable.

BAM’s team, led by guest researcher Gustav Graeber, has already shown that a liquid alkali-metal anode can deliver 100 times the power of graphite. Right now, though, that record output appears only at about 250°C (482°F). To bring the technology down to room temperature, the researchers add potassium to lower the anode’s melting point. Most solid electrolytes degrade in contact with potassium, so the electrolyte becomes the new bottleneck.

NASICON materials break that impasse. They conduct ions well at ambient conditions and tolerate potassium, especially when doped with hafnium. Hafnium, however, is scarce and expensive. The BAM project therefore, screens earth-abundant dopants that can match hafnium’s stabilizing effect. The most promising compositions are already being integrated and cycled in prototype sodium cells.

If the search succeeds, sodium-based solid-state batteries could move from the lab to everyday devices and electric vehicles. Higher energy density would extend operating time, while solid electrolytes would improve intrinsic safety. Faster charging and a supply chain that leans on plentiful sodium rather than scarce lithium and cobalt would make the technology attractive for grid storage as well—an incremental but meaningful step toward lower-carbon energy systems.

Saludos.

Vaya titular, parece que han dado con el (n-ésimo) Santo Grial de las baterías hasta que te lees la parte en rojo. Estos angloparlantes y sus "may"s en los titulares.

Yo tambien podría hacerme rico antes de que acabe el año.. si me tocara la lotería. Para eso tendría que empezar a jugar primero.. pero la posibilidad está ahí. "May"

[Cadavre Exquis]

Researchers Build 'The World's Fastest Petahertz Quantum Transistor'. They Predict Lightwave Electronics
Posted by EditorDavid on Saturday May 24, 2025 @12:34PM from the got-a-light dept.

"What if ultrafast pulses of light could operate computers at speeds a million times faster than today's best processors?" asks the University of Arizona.

"A team of scientists, including researchers from the University of Arizona, are working to make that possible."

In a groundbreaking international effort, researchers from the Department of Physics in the College of Science and the James C. Wyant College of Optical Sciences demonstrated a way to manipulate electrons in graphene using pulses of light that last less than a trillionth of a second. By leveraging a quantum effect known as tunneling, they recorded electrons bypassing a physical barrier almost instantaneously, a feat that redefines the potential limits of computer processing power. A study published in Nature Communications highlights how the technique could lead to processing speeds in the petahertz range — over 1,000 times faster than modern computer chips. Sending data at those speeds would revolutionize computing as we know it, said Mohammed Hassan, an associate professor of physics and optical sciences. Hassan has long pursued light-based computer technology and previously led efforts to develop the world's fastest electron microscope...

[T]he researchers used a laser that switches off and on at a rate of 638 attoseconds to create what Hassan called "the world's fastest petahertz quantum transistor... For reference, a single attosecond is one-quintillionth of a second," Hassan said. "That means that this achievement represents a big leap forward in the development of ultrafast computer technologies by realizing a petahertz-speed transistor." While some scientific advancements occur under strict conditions, including temperature and pressure, this new transistor performed in ambient conditions — opening the way to commercialization and use in everyday electronics. Hassan is working with Tech Launch Arizona, the office that works with investigators to commercialize inventions stemming from U of A research in order to patent and market innovations.

While the original invention used a specialized laser, the researchers are furthering development of a transistor compatible with commercially available equipment. "I hope we can collaborate with industry partners to realize this petahertz-speed transistor on a microchip," Hassan said.

Thanks to long-time Slashdot reader goslackware for sharing the news.

Saludos.

[Cadavre Exquis]

World-First Biocomputing Platform Hits the Market
Posted by BeauHD on Tuesday June 03, 2025 @11:30PM from the brain-in-a-vat dept.

An anonymous reader quotes a report from IEEE Spectrum:

In a development straight out of science fiction, Australian startup Cortical Labs has released what it calls the world's first code-deployable biological computer. The CL1, which debuted in March, fuses human brain cells on a silicon chip to process information via sub-millisecond electrical feedback loops. Designed as a tool for neuroscience and biotech research, the CL1 offers a new way to study how brain cells process and react to stimuli. Unlike conventional silicon-based systems, the hybrid platform uses live human neurons capable of adapting, learning, and responding to external inputs in real time. "On one view, [the CL1] could be regarded as the first commercially available biomimetic computer, the ultimate in neuromorphic computing that uses real neurons," says theoretical neuroscientist Karl Friston of University College London. "However, the real gift of this technology is not to computer science. Rather, it's an enabling technology that allows scientists to perform experiments on a little synthetic brain."

The first 115 units will begin shipping this summer at $35,000 each, or $20,000 when purchased in 30-unit server racks. Cortical Labs also offers a cloud-based "wetware-as-a-service" at $300 weekly per unit, unlocking remote access to its in-house cell cultures. Each CL1 contains 800,000 lab-grown human neurons, reprogrammed from the skin or blood samples of real adult donors. The cells remain viable for up to six months, fed by a life-support system that supplies nutrients, controls temperature, filters waste, and maintains fluid balance. Meanwhile, the neurons are firing and interpreting signals, adapting from each interaction.

The CL1's compact energy and hardware footprint could make it attractive for extended experiments. A rack of CL1 units consumes 850-1,000 watts, notably lower than the tens of kilowatts required by a data center setup running AI workloads. "Brain cells generate small electrical pulses to communicate to a broader network," says Cortical Labs Chief Scientific Officer Brett Kagan. "We can do something similar by inputting small electrical pulses representing bits of information, and then reading their responses. The CL1 does this in real time using simple code abstracted through multiple interacting layers of firmware and hardware. Sub-millisecond loops read information, act on it, and write new information into the cell culture."

The company sees CL1 as foundational for testing neuropsychiatric treatments, leveraging living cells to explore genetic and functional differences. "It allows people to study the effects of stimulation, drugs and synthetic lesions on how neuronal circuits learn and respond in a closed-loop setup, when the neuronal network is in reciprocal exchange with some simulated world," says theoretical neuroscientist Karl Friston of University College London. "In short, experimentalists now have at hand a little 'brain in a vat,' something philosophers have been dreaming about for decades."

Saludos.

[Cadavre Exquis]

Scientists in Japan Develop Plastic That Dissolves in Seawater Within Hours
Posted by msmash on Wednesday June 04, 2025 @01:30PM from the encouraging-feedback dept.

Researchers in Japan have developed a plastic that dissolves in seawater within hours, offering up a potential solution for a modern-day scourge polluting oceans and harming wildlife. From a report:

While scientists have long experimented with biodegradable plastics, researchers from the RIKEN Center for Emergent Matter Science and the University of Tokyo say their new material breaks down much more quickly and leaves no residual trace.

At a lab in Wako city near Tokyo, the team demonstrated a small piece of plastic vanishing in a container of salt water after it was stirred up for about an hour. While the team has not yet detailed any plans for commercialisation, project lead Takuzo Aida said their research has attracted significant interest, including from those in the packaging sector.

Saludos.

[Cadavre Exquis]

Samsung dry cell breakthrough set to bring cheaper solid-state batteries
Samsung's commercial solid-state battery launch is set to coincide with that of Toyota, as it tries to get an early start in the electric car market. It has now achieved a solid-state battery breakthrough that addresses the cost concerns.

Daniel Zlatev · 2025.06.17

Samsung's solid-state battery. (Image source: Marklines.com)

Samsung will be applying the dry production method that Tesla touted as a way to slash battery costs by half, to its upcoming solid-state batteries.

The goal is to lower the biggest hurdle before mass solid-state battery adoption for electric vehicles, their manufacturing costs, as Samsung aims to launch them in 2027.

Coincidentally, that is when both Toyota and the world's largest battery maker CATL said they will start mass solid-state battery production, too.

If Samsung masters the dry electrode production method that Tesla is now scaling for cheaper manufacturing of the Cybertruck's 4680 batteries, it may realize cost advantages and flip the script on the Chinese juggernauts when the superior battery chemistry starts being installed in electric vehicles en masse.

Samsung's solid-state battery specs are already some of the best in the industry, hitting the technology's 500 Wh/kg energy density potential. Due to the inherent advantages of the solid-state battery technology, Samsung teased a 9-minute charging time and 600 miles on a charge from a pack with the footprint of current EV batteries.

What's even more important, though, Samsung is focusing on solid-state battery production costs, something that initially made CATL peg them for mass release no earlier than 2030. The biggest battery maker has come around since then, acknowledging how fast the field is progressing and saying that its own solid-state battery program will be ready for mass production in 2027.

Samsung is banking on two breakthrough production methods to lower its solid-state battery costs. The first is roll pressing, a procedure that doesn't require cumbersome sealing of the cell with the so-called Warm Istactic Press technique prior to applying 600 MPa high-temperature pressure underwater to fuse the electrode and electrolyte materials into a solid.

Now Samsung has detailed a fiberization process that it is bringing not only to its pilot solid-state battery line production, but also to its conventional batteries in order to make them much cheaper to produce.

Samsung pegs the binder as the chief difference between the more energy-efficient dry electrode method, and the common wet production that includes coating with the use of toxic solvents, and baking in huge furnaces to dry up afterward.

Samsung's Teflon binder is calculated to stretch under the force and pressure of the roll pressing method, forming a "layer that supports the conductive material and the active material." The resulting separation film is thus stronger and evenly distributed, preventing the electrodes and solid electrolyte from coming in direct contact, while still allowing the free flow of charge between them.

Tesla is doing something similar with its dry-cathode battery production method, but has yet to manufacture such cells at scale to replace the second generation 4680 battery with 15% higher energy density that it uses in the Cybertruck on the cheap.

Saludos.

[Cadavre Exquis]

UK Scientists Achieve First Commercial Tritium Production
Posted by EditorDavid on Saturday July 05, 2025 @06:34PM from the breeding-success dept.

Interesting Engineering reports:

Astral Systems, a UK-based private commercial fusion company, has claimed to have become the first firm to successfully breed tritium, a vital fusion fuel, using its own operational fusion reactor. This achievement, made with the University of Bristol, addresses a significant hurdle in the development of fusion energy....

Scientists from Astral Systems and the University of Bristol produced and detected tritium in real-time from an experimental lithium breeder blanket within Astral's multi-state fusion reactors. "There's a global race to find new ways to develop more tritium than what exists in today's world — a huge barrier is bringing fusion energy to reality," said Talmon Firestone, CEO and co-founder of Astral Systems. "This collaboration with the University of Bristol marks a leap forward in the search for viable, greater-than-replacement tritium breeding technologies. Using our multi-state fusion technology, we are the first private fusion company to use our reactors as a neutron source to produce fusion fuel."

Astral Systems' approach uses its Multi-State Fusion (MSF) technology. The company states this will commercialize fusion power with better performance, efficiency, and lower costs than traditional reactors. Their reactor design, the result of 25 years of engineering and over 15 years of runtime, incorporates recent understandings of stellar physics. A core innovation is lattice confinement fusion (LCF), a concept first discovered by NASA in 2020. This allows Astral's reactor to achieve solid-state fuel densities 400 million times higher than those in plasma. The company's reactors are designed to induce two distinct fusion reactions simultaneously from a single power input, with fusion occurring in both plasma and a solid-state lattice.

The article includes this quote from professor Tom Scott, who led the University of Bristol's team, supported by the Royal Academy of Engineering and UK Atomic Energy Authority. "This landmark moment clearly demonstrates a potential path to scalable tritium production in the future and the capability of Multi-State Fusion to produce isotopes in general."

And there's also this prediction from the company's web site:

"As we progress the fusion rate of our technology, aiming to exceed 10 trillion DT fusions per second per system, we unlock a wide range of applications and capabilities, such as large-scale medical isotope production, fusion neutron materials damage testing, transmutation of existing nuclear waste stores, space applications, hybrid fusion-fission power systems, and beyond."

"Scientists everywhere are racing to develop this practically limitless form of energy," write a climate news site called The Cooldown. (Since in theory nuclear fusion "has an energy output four times higher than that of fission, according to the International Atomic Energy Agency.")

Thanks to long-time Slashdot reader fahrbot-bot for sharing the news.

Saludos.

[Cadavre Exquis]

Tesla Launches Solar-Powered 'Oasis' Supercharger Station: 30-Acre Solar Farm, 39 MWh of Off-Grid Batteries
Posted by EditorDavid on Sunday July 06, 2025 @03:34AM from the powering-ahead dept.

"Tesla has launched its new Oasis Supercharger," reports Electrek, "the long-promised EV charging station of the future, with a solar farm and off-grid batteries."

Early in the deployment of the Supercharger network, Tesla promised to add solar arrays and batteries to the Supercharger stations, and CEO Elon Musk even said that most stations would be able to operate off-grid... Last year, Tesla announced a new project called 'Oasis', which consists of a new model Supercharger station with a solar farm and battery storage enabling off-grid operations in Lost Hills, California.

Tesla has now unveiled the project and turned on most of the Supercharger stalls. The project consists of 168 chargers, with half of them currently operational, making it one of the largest Supercharger stations in the world. However, that's not even the most notable aspect of it. The station is equipped with 11 MW of ground-mounted solar panels and canopies, spanning 30 acres of land, and 10 Tesla Megapacks with a total energy storage capacity of 39 MWh. It can be operated off-grid, which is the case right now, according to Tesla.

With off-grid operations, Tesla was about to bring 84 stalls online just in time for the Fourth of July travel weekend. The rest of the stalls and a lounge are going to open later this year.

The article makes that point that "This is what charging stations should be like: fully powered by renewable energy."

Saludos.

[Cadavre Exquis]

Google DeepMind's Spinoff Company 'Very Close' to Human Trials for Its AI-Designed Drugs
Posted by EditorDavid on Monday July 07, 2025 @12:36AM from the AI-on-drugs dept.

Google DeepMind's chief business officer says Alphabet's drug-discovery company Isomorphic Labs "is preparing to launch human trials of AI-designed drugs," according to a report in Fortune, "pairing cutting-edge AI with pharma veterans to design medicines faster, cheaper, and more accurately."

"There are people sitting in our office in King's Cross, London, working, and collaborating with AI to design drugs for cancer," said Colin Murdoch [DeepMind's chief business officer and president of Isomorphic Labs]. "That's happening right now."

After years in development, Murdoch says human clinical trials for Isomorphic's AI-assisted drugs are finally in sight. "The next big milestone is actually going out to clinical trials, starting to put these things into human beings," he said. "We're staffing up now. We're getting very close."

The company, which was spun out of DeepMind in 2021, was born from one of DeepMind's most celebrated breakthroughs, AlphaFold, an AI system capable of predicting protein structures with a high level of accuracy. Interactions of AlphaFold progressed from being able to accurately predict individual protein structures to modeling how proteins interact with other molecules like DNA and drugs. These leaps made it far more useful for drug discovery, helping researchers design medicines faster and more precisely, turning the tool into a launchpad for a much larger ambition... In 2024, the same year it released AlphaFold 3, Isomorphic signed major research collaborations with pharma companies Novartis and Eli Lilly. A year later, in April 2025, Isomorphic Labs raised $600 million in its first-ever external funding round, led by Thrive Capital. The deals are part of Isomorphic's plan to build a "world-class drug design engine..."

Today, pharma companies often spend millions attempting to bring a single drug to market, sometimes with just a 10% chance of success once trials begin. Murdoch believes Isomorphic's tech could radically improve those odds. "We're trying to do all these things: speed them up, reduce the cost, but also really improve the chance that we can be successful," he says. He wants to harness AlphaFold's technology to get to a point where researchers have 100% conviction that the drugs they are developing are going to work in human trials. "One day we hope to be able to say — well, here's a disease, and then click a button and out pops the design for a drug to address that disease," Murdoch said. "All powered by these amazing AI tools."

Saludos.

[Cadavre Exquis]

COVID-19 Vaccine's mRNA Technology Adapted for First Antibiotic-Resistant Bacteria Vaccine
Posted by EditorDavid on Monday July 14, 2025 @07:34AM from the giving-it-a-shot dept.

Researchers have created the world's first mRNA-based vaccine against a deadly, antibiotic-resistant bacterium — and they did it using the platform developed for COVID-19 vaccines.

Medical Express publishes their announcement:

The vaccine developed by the team from the Institute for Biological Research and Tel Aviv University is an mRNA-based vaccine delivered via lipid nanoparticles, similar to the COVID-19 vaccine. However, mRNA vaccines are typically effective against viruses like COVID-19 — not against bacteria like the plague... In 2023, the researchers developed a unique method for producing the bacterial protein within a human cell in a way that prompts the immune system to recognize it as a genuine bacterial protein and thus learn to defend against it.

The researchers from Tel Aviv University and the Institute for Biological Research proved, for the first time, that it is possible to develop an effective mRNA vaccine against bacteria. They chose Yersinia pestis, the bacterium that causes bubonic plague — a disease responsible for deadly pandemics throughout human history. In animal models, the researchers demonstrated that it is possible to effectively vaccinate against the disease with a single dose.

The team of researchers was led by Professor Dan Peer at Tel Aviv University, a global pioneer in mRNA drug development, who says the success of the current study now "paves the way for a whole world of mRNA-based vaccines against other deadly bacteria."

Saludos.

[Cadavre Exquis]

https://x.com/alexwei_/status/1946477742855532918

OpenAI's gold medal performance on the International Math Olympiad

This feels notable to me. OpenAI research scientist Alexander Wei:

I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).

We evaluated our models on the 2025 IMO problems under the same rules as human contestants: two 4.5 hour exam sessions, no tools or internet, reading the official problem statements, and writing natural language proofs. [...]

Besides the result itself, I am excited about our approach: We reach this capability level not via narrow, task-specific methodology, but by breaking new ground in general-purpose reinforcement learning and test-time compute scaling.

In our evaluation, the model solved 5 of the 6 problems on the 2025 IMO. For each problem, three former IMO medalists independently graded the model’s submitted proof, with scores finalized after unanimous consensus. The model earned 35/42 points in total, enough for gold!

HUGE congratulations to the team—Sheryl Hsu, Noam Brown, and the many giants whose shoulders we stood on—for turning this crazy dream into reality! I am lucky I get to spend late nights and early mornings working alongside the very best.

Btw, we are releasing GPT-5 soon, and we’re excited for you to try it. But just to be clear: the IMO gold LLM is an experimental research model. We don’t plan to release anything with this level of math capability for several months.

Here's Wikipedia on the International Mathematical Olympiad:

It is widely regarded as the most prestigious mathematical competition in the world. The first IMO was held in Romania in 1959. It has since been held annually, except in 1980. More than 100 countries participate. Each country sends a team of up to six students, plus one team leader, one deputy leader, and observers.

This year's event is in Sunshine Coast, Australia. Here's the web page for the event, which includes a button you can click to access a PDF of the six questions - maybe they don't link to that document directly to discourage it from being indexed.

The first of the six questions looks like this:

Alexander shared the proofs produced by the model on GitHub. They're in a slightly strange format - not quite MathML embedded in Markdown - which Alexander excuses since "it is very much an experimental model".

The most notable thing about this is that the unnamed model achieved this score without using any tools. OpenAI's Sebastien Bubeck emphasizes that here:

Just to spell it out as clearly as possible: a next-word prediction machine (because that's really what it is here, no tools no nothing) just produced genuinely creative proofs for hard, novel math problems at a level reached only by an elite handful of pre‑college prodigies.

There's a bunch more useful context in this thread by Noam Brown, including a note that this model wasn't trained specifically for IMO problems:

Typically for these AI results, like in Go/Dota/Poker/Diplomacy, researchers spend years making an AI that masters one narrow domain and does little else. But this isn’t an IMO-specific model. It’s a reasoning LLM that incorporates new experimental general-purpose techniques.

So what’s different? We developed new techniques that make LLMs a lot better at hard-to-verify tasks. IMO problems were the perfect challenge for this: proofs are pages long and take experts hours to grade. Compare that to AIME, where answers are simply an integer from 0 to 999.

Also this model thinks for a long time. o1 thought for seconds. Deep Research for minutes. This one thinks for hours. Importantly, it’s also more efficient with its thinking. And there’s a lot of room to push the test-time compute and efficiency further.

It’s worth reflecting on just how fast AI progress has been, especially in math. In 2024, AI labs were using grade school math (GSM8K) as an eval in their model releases. Since then, we’ve saturated the (high school) MATH benchmark, then AIME, and now are at IMO gold. [...]

When you work at a frontier lab, you usually know where frontier capabilities are months before anyone else. But this result is brand new, using recently developed techniques. It was a surprise even to many researchers at OpenAI. Today, everyone gets to see where the frontier is.

Saludos.

P.D. Por si quieren ver como resolvería el problema un humano:

https://x.com/wtgowers/status/1946246810923589887

Timothy Gowers · 2025.07.18 · IMO2025Q1

[Cadavre Exquis]

https://www.pressreader.com/spain/Cinco-Dias/20250731/page/32/textview

ITER, el mayor experiment­o energético del mundo

Saludos.

[Cadavre Exquis]

https://www.expansion.com/directivos/2025/07/31/688a4cc3e5fdea2a748b459b.html

Saludos.

[Cadavre Exquis]

La vacuna para prevenir el contagio por VIH va a tener genéricos libres de licencia para su uso en países de alta incidencia y recursos limitados

Por @Wicho — 28 de julio de 2025

¡Vacunas FTW!

Hace unos meses hablábamos de que Lenacapavir, un medicamento inyectable que protege prácticamente al 100 % contra la infección por VIH con una dosis cada seis meses, había sido escogido como avance científico del año por la revista Nature. Hoy me acabo de enterar de que su uso para prevención del VIH, imagino que muy a pesar de Robert Kennedy Jr., ha sido autorizado en los Estados Unidos. Y de que, además, se van a fabricar genéricos libres de licencia.

Del artículo de Science:

Un extenso ensayo de eficacia en adolescentes y mujeres jóvenes africanas reveló en junio que estas inyecciones reducían a cero las infecciones por el VIH, con una asombrosa eficacia del 100%. Cualquier duda sobre el hallazgo desapareció tres meses después, cuando un ensayo similar, realizado en cuatro continentes, informó de una eficacia del 99,9% en personas de distinto sexo que practican sexo con hombres.

Gilead, el fabricante, que la va a comercializar para este uso con el nombre Yeztugo, está tramitando también la aprobación para su uso en otros países como Australia, Brasil, Canadá, Sudáfrica y Suiza. Y está preparando solicitudes para Argentina, México y Perú.

En Europa ha presentado una solicitud de autorización de comercialización y una solicitud EU-M4all ante la Agencia Europea de Medicamentos (EMA), ambas aceptadas para evaluación bajo el proceso acelerado. Con la solicitud EU-M4all la EMA, en cooperación con la Organización Mundial de la Salud (OMS), puede emitir dictámenes científicos sobre medicamentos de uso humano de alta prioridad, incluidas las vacunas, destinados a mercados fuera de la Unión Europea (UE), lo que facilitará su uso en otros mercados.

Gilead también ha anunciado un acuerdo con el Fondo Mundial para la lucha contra el sida, la tuberculosis y la malaria para suministrar dosis suficientes del medicamento para tratar hasta a dos millones de personas durante tres años en los países en los que actúa el Fondo Mundial, sin beneficio alguno para la empresa farmacéutica.

Y además habrá genéricos libres de licencia del medicamento para su uso en 120 países de alta incidencia y recursos limitados.

Aunque todo esto viene con un pequeño jarro de agua fría, pues como me cuenta Belén Tarrafeta, que de esto sabe un rato largo, esos genéricos no estarán disponibles hasta 2027. Y hay muchos países excluidos de la lista donde se podrán comercializar, incluida gran parte de América Latina.

Belén añade también que en los contratos actuales del Fondo Mundial con Gilead, el precio es secreto. Y si no sabes el precio, difícilmente puedes planificar nada para los programas de salud pública.

Pero de todas formas quiero quedarme con una visión optimista de esta noticia. Hace 40 años una infección por VIH era una sentencia de muerte. Unas décadas después, gracias a los tratamientos con antirretrovirales, entre ellos el Lenacapavir, de hecho, se convirtió en una enfermedad crónica tratable. Y ahora podemos prevenir la infección.

Lo de los tratamientos y la vacuna siempre que quien los necesite tenga acceso a ellos, claro. Y en esto hay que trabajar. Pero al menos ahora tenemos esa esperanza. Para quienes en los 80 ya teníamos uso de razón esta noticia es una de las más maravillosas e increíbles que podemos haber vivido. Y que durante mucho tiempo pareció que nunca llegaría.

_____
Foto de Diana Polekhina en Unsplash

Saludos.