Topic
Explore the latest artificial intelligence news and research.
Xiaomi's second-generation SoC, the XRING O3, stays on 3nm while expanding to 10 high-performance CPU cores, a 16-core GPU, LPDDR6, and a 200TOPS NPU. This article separates official figures from measurements on a development board to examine what the 113.8GB/s memory bandwidth and 20W-class peak power actually mean.
d-Matrix's Raptor accelerator measured roughly 105 TB/s of card-level bandwidth using 3D-DRAM in real silicon. But its inference speedups come from architectural modeling, not third-party benchmarks, and production yield data remains unpublished.
The anonymous model Ox Alpha has different data retention terms depending on the access route, and even within OpenRouter, the model description and Stealth Program terms conflict. Here's why companies should evaluate performance and data contracts separately.
Cursor's Git infrastructure treats an S3-based write-ahead log as the source of truth, with local NVMe serving as a regenerable cache. We examine the design's strengths and open questions as it consolidates write ordering into S3 CAS and decouples replica count from push latency.
A peer-reviewed study testing 10 AI models found that majority-following behavior in AI agents varies with group size and model. The 1,000-agent figure is not evidence of complex cooperation but the upper limit tested in a binary-choice experiment.
Five US agencies have flagged an active threat involving AI-generated Python scripts targeting internet-exposed Siemens S7 PLCs. The activity is likely reconnaissance, with agencies urging exposure reduction and monitoring for unusual read/write access.
Edgewater Research has suggested a multi-year memory contract between NVIDIA and Micron. While the companies' mass-production relationship and Micron's anonymous strategic customer agreements are already public, the contract's duration, volume, and pricing remain unconfirmed.
Alibaba is accelerating its AI infrastructure investment and aims to shorten the payback period to 2.5 years by expanding the use of self-developed chips. Amid U.S. export restrictions, the company is strengthening its own silicon foundation to boost profit margins, though winning external customers and navigating international regulations remain key challenges ahead.
An ad-creation system combining generative AI with active learning outperformed human-made ads in average Instagram link CTR. However, its effect on bookings or sales remains unverified.
NVIDIA says Claude Opus 5, running on its proprietary AVO agent framework, achieved a perfect score on the public ARC-AGI-3 benchmark, versus roughly 30% for the model alone under different measurement conditions. We examine how this points to a structural shift in which harness design—not the underlying model—increasingly determines real-world AI performance, and why it also serves NVIDIA's push to expand GPU demand.
