Toggle light / dark theme

Humans beat AI at international math contest despite gold-level AI scores

Humans beat generative AI models made by Google and OpenAI at a top international mathematics competition, despite the programs reaching gold-level scores for the first time.

Neither model scored full marks—unlike five young people at the International Mathematical Olympiad (IMO), a prestigious annual competition where participants must be under 20 years old.

Google said Monday that an advanced version of its Gemini chatbot had solved five out of the six set at the IMO, held in Australia’s Queensland this month.

Mathematical model sheds light on internal ocean waves and climate prediction

Deep below the surface of the ocean, unseen waves roil and churn the water. These internal waves, traveling between water layers of different temperatures and densities, draw cold, nutrient-rich water up from the depths and play a major role in oceanic circulation. Understanding and modeling their behavior is critical for developing more accurate simulations of an increasingly unpredictable climate.

In a Nature Communications paper, Rensselaer Polytechnic Institute (RPI) Math Professor Yuri V. Lvov, Ph.D. and a team of oceanographers develop a first-of-its-kind model of internal wave dynamics that lays the foundation for new, more reliable models of ocean circulation.

“Internal, wave-driven, vertical mixing is believed to be a main driver of oceanic circulation,” Lvov said. “It shapes Earth’s climate by influencing sea level rise, nutrient fluxes, , and anthropogenic heat and carbon uptake.”

Scientists advance efforts to create ‘virtual cell lab’ as testing ground for future research with live cells

Using mathematical analysis of patterns of human and animal cell behavior, scientists say they have developed a computer program that mimics the behavior of such cells in any part of the body. Led by investigators at Indiana University, Johns Hopkins Medicine, the University of Maryland School of Medicine and Oregon Health & Science University, the new work was designed to advance ways of testing and predicting biological processes, drug responses and other cell dynamics before undertaking more costly experiments with live cells.

With further work on the program, the researchers say it could eventually serve as a “digital twin” for testing any drug’s effect on cancer or other conditions, gene environment interactions during brain development, or any number of dynamic cellular molecular processes in people where such studies are not possible.

The new study and examples of cell simulations are described online July 25 in the journal Cell.

NVIDIA Brings Reasoning Models to Consumers Ranging from 1.5B to 32B Parameters

Today, NVIDIA unveiled OpenReasoning-Nemotron, a quartet of distilled reasoning models with 1.5B, 7B, 14B, and 32B parameters, all derived from the 671B-parameter DeepSeek R1 0528. By compressing that massive teacher into four leaner Qwen‑2.5-based students, NVIDIA is making advanced reasoning experiments accessible even on standard gaming rigs, without the need to worry about hefty GPU bills and cloud usage. The key is not some elaborate trick but raw data. Using the NeMo Skills pipeline, NVIDIA generated five million math, science, and code solutions, and then fine-tuned each one purely with supervised learning. Already, the 32B model hits an 89.2 on AIME24 and 73.8 on the HMMT February contest, while even the 1.5B variant manages a solid 55.5 and 31.5.

Nima Arkani-Hamed, Gopal Prasad Professor, School of Natural Sciences, Institute for Advanced Study

Beyond Space-Time and Quantum Mechanics.

Nima Arkani-Hamed.

(June 28, 2025)


A tribute to jim simons in celebration of the importance of basic science and mathematics.

Leaders in mathematics, science and philanthropy gathered on June 27, 2025, to remember the incredible contributions of Jim Simons and to inspire continued philanthropic support of basic research.

This Rope-Powered Robot Dog Built by a US Student Walks With Stunning Realism Thanks to a Brilliant Mathematical Design

IN A NUTSHELL 🐕 CARA is a robot dog created by a Purdue University student using innovative capstan drive technology. 🔧 The robot incorporates custom 3D-printed parts and high-strength materials like carbon fiber for durability and efficiency. 🤖 Advanced coding techniques such as Inverse Kinematics allow CARA to move with natural grace and agility. 🚀

/* */