A quieter feed day — no new Innermost Loop or NVIDIA blog posts in the window — but a clear theme: decision models (outputs that score actions, not chat). Microsoft shipped a Qwen-based Decision 1, startups open-sourced compact scorers, and opinion pieces questioned whether vanilla LLMs “decide” at all.

Today

Watch: Open-weight models · Mistral

Built with utilities/ai_report_build.py (24-hour lookback). No Innermost Loop or NVIDIA posts in this window. Edited for publication. About the AI Report.