Nvidia's Nemotron-Cascade 2 wins math and coding gold medals with 3B active parameters — and its post-training recipe is now open-source

Nvidia's Nemotron-Cascade 2 is a 30B MoE model that activates only 3B parameters at inference time, yet achieved gold ...
VentureBeat

New open-source math model Light-R1-32B surpasses equivalent DeepSeek performance with only $1000 in training costs

Researchers have introduced Light-R1-32B, a new open-source AI model optimized to solve advanced math problems. It is now available on Hugging Face under a permissive Apache 2.0 license — free for ...
Nanowerk

New math model tackles current crowding in 2D thin-film nanoelectronics

The new framework replaces approximations, accurately modeling current crowding and spreading resistance in 2D materials to ...
TechCrunch

DeepSeek upgrades its math-focused AI model Prover

Chinese AI lab DeepSeek has quietly updated Prover, its AI model that’s designed to solve math-related proofs and theorems. According to South China Morning Post, DeepSeek uploaded the latest version ...
The Verge

Microsoft’s small math AI model does math better than the big boys.

Microsoft found that small language models can exceed the performance of much larger ones when trained to specialize in a single area. Researchers fine-tuned the Mistral 7B model to create Orca-Math, ...
Yahoo

OpenAI Model Earns Gold-Medal Score at International Math Olympiad and Advances Path to Artificial General Intelligence

Add Yahoo as a preferred source to see more of our stories on Google. A few months before the 2025 International Mathematical Olympiad (IMO) in July, a three-person team at OpenAI made a long bet that ...