AI Intel

vLLM released version v0.31.1rc0, introducing a new metrics feature to expose cached prompt tokens by cache tier, aiding performance monitoring.

A cyberattack on several major South Korean banks is suspected to have been carried out by a single individual utilizing a combination of open-source…

Semwright is a new product designed to provide AI agents with structured access to real-world software applications.

OpenSEO is presented as an open-source alternative to Semrush, offering tools for SEO analysis and optimization.

A user conducted local performance tests on an RTX 4090 GPU to compare the speeds of several new open decision models, including laya, liquid's d1, c…

The jevman project showcases AI decision models playing Pac-Man, comparing the performance of six popular decision models, including OpenAI's decisio…

A $2800 rig featuring 8x Radeon Pro V620 GPUs (256GB VRAM) and a custom vLLM fork can run Qwen3.8-Flash-Next at 60-100 t/s decode and over 3000 t/s p…

A new arXiv paper introduces an "Adaptive Workflow Intelligence" cognitive architecture, aiming to address the brittleness of AI-driven enterprise au…

A new empirical study investigates the downstream utility of agent skills, highlighting that a relevant skill does not necessarily translate to impro…

A new arXiv paper introduces KVFetch, a temporal prefetching technique for KV cache compression, essential for efficient LLM inference with large con…