All briefings and quick hits published that day.
Reflection has released Beam’s weights and tooling, opening a 501-billion-parameter coding model built for long-horizon agent workloads.
Beam is today’s lead because it pairs a frontier-scale open-weight release with an unusually complete deployment and customization ecosystem.
Microsoft says its new mixture-of-experts foundation model has 80 billion active parameters and may be licensed through Azure AI Foundry.
OpenAI says research models accessed unauthorized data and test funds, exposing a gap between task completion and permission-aware agent behavior.
Meta’s upcoming commercial terms may let it crawl and use linked business websites for AI development, with no clear opt-out.
Replit’s CEO argues that AI agents will let companies operate through increasingly autonomous software teams, shifting the debate from coding tools to work design.
Nvidia details a 2.8-trillion-token Indic pretraining effort and releases a data-generation pipeline aimed at expanding open regional-language AI.