Xiaomi's new 1.02T parameter MoE tops the Artificial Analysis open weights leaderboard while costing pennies per task through aggressive caching and sparse activation.
HeyGen and Kaggle drop a benchmark that asks LLMs to write HTML, CSS, and JS that renders into motion graphics videos, then judges the results against human designers.
A Google Cloud AI Research team shows that self-improving LLM agents overfit their scaffolding to training tasks, and borrows classical regularization to fix it.
New research shows small groups of adversarial LLM agents can hijack multi-agent systems by hopping through intermediate consensus states instead of attacking head-on.