Mistral’s new Large 4 trails some Chinese open models in independent tests

IT Hardware

AI-rewritten: This is a summary of an article from Tom’s Hardware, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

Tom’s Hardware • Shane Downing • October 9, 2026

Mistral recently launched its Large 4 (ML4) model, which uses a mixture-of-experts architecture trained on Nvidia’s Grace Blackwell GPUs in European datacenters. Artificial Analysis (AA), an independent benchmarking firm, ranked ML4 as the most intelligent open-weight model outside the U.S. and China with a score of 38 on its Intelligence Index v4.3.2. However, five Chinese models scored higher overall, including MiMo-V2.6-Pro at 46 and GLM-5.3 (max) at 45. Unlike these competitors, ML4 is not yet open; Mistral states its weights will be released by the end of October.

READ MORE

PC shipments tumble over 20% in 3Q26 as chip shortages bite

IT Hardware

AI-rewritten: This is a summary of an article from Tom’s Hardware, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

Tom’s Hardware • Jowi Morales • October 9, 2026

Global PC shipments fell by 20.1% year-over-year in the third quarter of 2026, representing a drop of 15.8 million units. According to IDC Global, this decline is steeper than the 3.8% decrease seen last quarter and down 9.1% from the second quarter of this year. Many research firms had warned about this trend since late last year, with IDC predicting the market could shrink up to 9%, particularly in the budget sector. Some experts estimate the situation will not improve until 2029, while Tom’s Hardware readers stated they do not plan to upgrade their systems in the next two years.

READ MORE

Liquid-crystal-like magnetism explains puzzling properties in a rare-earth compound

Quantum Technology

AI-rewritten: This is a summary of an article from Phys.org, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

Phys.org • October 8, 2026

Researchers led by Pengcheng Dai at Rice University discovered that quantum spins in the rare-earth compound YbMnBi₂ behave like molecules in a liquid crystal. In this state, spins favor a specific direction without forming large-scale magnetic order. This phenomenon was observed when the team fired neutron beams at crystals of YbMnBi₂ and a related calcium compound to measure spin fluctuations under varying temperatures and magnetic fields.

READ MORE

Layered semiconductor unlocks magnetic control of light emitted by quantum condensates

Quantum Technology

AI-rewritten: This is a summary of an article from Phys.org, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

Phys.org • October 8, 2026

Researchers have developed a method to control light emitted by quantum condensates using the magnetic properties of a new layered semiconductor called chromium sulfide bromide (CrSBr). Unlike traditional semiconductors that require extreme cooling, this material allows exciton–polaritons to form at moderate temperatures. Exciton–polaritons are hybrid states created when electron-hole pairs, known as excitons, couple with light in an optical resonator. This coupling reduces their effective mass, enabling them to move in lockstep and form a macroscopic quantum state that emits coherent light.

READ MORE

How researchers tell different quantum excitations apart in individual molecules

Quantum Technology

AI-rewritten: This is a summary of an article from Phys.org, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

Phys.org • October 8, 2026

Dr. Arnab Banerjee and his colleagues faced a challenge when studying individual cobaltocene molecules using tunneling spectroscopy. The measured spectra displayed confusing signals caused by various excitations, making it difficult to determine which specific process created each measurement signal because theoretical models were unreliable. To solve this, the team placed the molecules on a lead surface that becomes superconducting at very low temperatures and used a scanning tunneling microscope with a metallic tip positioned extremely close to the molecule to allow electrons to cross the tiny gap.

READ MORE

Superconducting circuit links smaller photon groups into larger entangled states

Quantum Technology

AI-rewritten: This is a summary of an article from Phys.org, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

Phys.org • October 9, 2026

Researchers at Tsinghua University and the Hefei National Laboratory recently introduced a new strategy to link smaller groups of microwave photons into larger entangled states. This work, published in *Nature Physics*, aims to create adjustable graph states, which are mathematical networks describing connections between qubits used in quantum computing. The team utilized superconducting circuits, materials that allow electrical current to flow with zero resistance at specific low temperatures, to generate these photon groups.

READ MORE

Roundtables: A Conversation With the Creator of AI-Designed Viruses

Artificial Intelligence

AI-rewritten: This is a summary of an article from MIT Technology Review, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

MIT Technology Review • MIT Technology Review • October 9, 2026

Stanford University PhD student Samuel King recently used a generative AI model to propose genetic blueprints for microscopic viruses. While this is not yet an example of AI-generated life, it suggests that creating new life forms could be the next step. James O’Donnell, senior AI reporter, will interview King on October 16th at 18:30 BST regarding these findings and his recognition as one of MIT Technology Review’s Innovators Under 35.

READ MORE

We’re still figuring out the side effects of GLP-1 weight-loss drugs

Artificial Intelligence

AI-rewritten: This is a summary of an article from MIT Technology Review, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

MIT Technology Review • Jessica Hamzelou • October 9, 2026

Research presented at an aging meeting in Boston suggests that overweight and diabetic people taking GLP-1 weight-loss drugs have a biological age two to three years younger than those on a placebo. Eli Lilly and Novo Nordisk, the companies that make these drugs, provided these findings. While approved for diabetes, kidney disease, and obesity, these medications are also being investigated for substance-use disorders, neurodegenerative disease, and cancer. A recent Gallup poll indicates 11% of Americans currently take a GLP-1 drug for weight loss, up from 3% in 2024, with 15% having used one at some point.

READ MORE

Job titles of the future: Delivery drone air traffic controller

Artificial Intelligence

AI-rewritten: This is a summary of an article from MIT Technology Review, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

MIT Technology Review • Eamon Whalen • October 9, 2026

Trevor Wischnewsky is now a remote pilot in command at FlyTrex, a role similar to an air traffic controller but for delivery drones. He manages fleets of hexacopter drones that transport cargo within a three-mile radius of Dallas. His job involves coordinating traffic with swarms from competitors like Zipline, Amazon, and Alphabet’s Wing using inputs from GPS, lasers, and sensors. To work there, Wischnewsky needed an FAA drone pilot certification and completed weeks of FAA-approved training.

READ MORE

We’re putting too much faith in AI’s ability to say no

Artificial Intelligence

AI-rewritten: This is a summary of an article from MIT Technology Review, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

MIT Technology Review • Arthur Holland Michel • October 9, 2026

People often assume artificial intelligence can naturally refuse harmful requests, but early models lacked this ability. Steven Adler of OpenAI noted that initial versions would answer any question, including how to kill oneself with a gun. Ryan McBain of Harvard added that these early systems could easily generate violent instructions because they were trained on vast amounts of web content containing violence and vitriol.

READ MORE