InfoWorld

Multi-token prediction technique triples LLM inference speed without auxiliary draft models

With reported 3x speed gains and limited degradation in output quality, the method targets one of the biggest pain points in production AI systems: latency at scale.
TMCnet

Inception Launches Mercury 2, the Fastest Reasoning LLM - 5x Faster Than Leading Speed-Optimized LLMs, with Dramatically Lower Inference Cost

Inception, the company behind the first commercial diffusion large language models (dLLMs), today announced the launch of Mercury 2, the fastest reasoning LLM and first reasoning dLLM. Mercury 2 ...
Psychology Today

Can LLMs Think Like Us?

In the complexity of human cognition, the hippocampus stands as a central player, orchestrating more than just the storage of memories. It is a master of inference—a cognitive ability that allows us ...

Researchers baked 3x inference speedups directly into LLM weights — without speculative decoding

Researchers from the University of Maryland, Lawrence Livermore, Columbia and TogetherAI have developed a training technique that triples LLM inference speed without auxiliary models or infrastructure ...
Opinion
CIOOpinion

AI isn’t failing, people are failing with AI

AI isn’t the problem — rushing it into the wrong tasks without the right data, expertise or guardrails is what makes projects fall apart.
Geeky Gadgets

Meta’s Vision-Language Shift VL-JEPA Beats Bulky LLMs

What if the AI systems we rely on today, those massive, resource-hungry large language models (LLMs)—were on the brink of being completely outclassed? Better Stack walks through how Meta’s VL-JEPA, a ...