Google Cloud unveils OpenRL, llm-d 1.0, and Agent Substrate at KubeCon 2026
Google Cloud at KubeCon 2026 released OpenRL, llm-d 1.0, and Agent Substrate — an open stack for RL fine-tuning, distributed inference, and safe AI agents on Kubernetes.
AI news and practical stories for this audience.
Google Cloud at KubeCon 2026 released OpenRL, llm-d 1.0, and Agent Substrate — an open stack for RL fine-tuning, distributed inference, and safe AI agents on Kubernetes.
On October 15, 2026, Meta released Muse Spark Open, an open-weight MoE model with 816B total parameters and 42B active per token; it accepts text, images, audio, and video and generates text only.
On October 16, 2026, Reflection released Beam, a 501B-parameter mixture-of-experts model under Apache 2.0, with 23B active parameters per token and a 1M context window.
StudentSim models individual learner responses and reactions to tutor guidance. It is a research system, not a classroom product.
A machine-learning system estimates space-weather risks at 66,935 substations in the continental US, producing location-specific forecasts 30–60 minutes before potential impacts.
Anthropic reports that Claude, working largely autonomously, wrote a complete formal Lean proof of Fermat's Last Theorem. The proof is public on GitHub for outside checking.