Mastering Long Contexts in LLMs with KVPress

Granite 4.2 LLMs: How They're Built

LLMs help robots understand vague instructions and focus on key details

To help robots do chores in places like homes and factories, a new approach from MIT uses one language model to clarify users’ instructions, then another to ignore irrelevant info.

Thinking to recall: How reasoning unlocks parametric knowledge in LLMs

Generative AI

GLM-5.2: Built for Long-Horizon Tasks