Scaling laws revisited: a closer look at compute-optimal inference 2026-01-22 · arxiv.org/abs/2601.12345 scaling-lawsinference
Anthropic publishes a constitution for agentic assistants 2026-01-18 · anthropic.com/news/constitution agentssafety
OpenAI releases structured-output guarantees for o-series models 2026-01-12 · openai.com/index/structured tool-useinfrastructure
A survey of mechanistic interpretability at the 2025 frontier 2026-01-05 · arxiv.org/abs/2601.00042 interpretabilitysurvey
RAG is mostly a retrieval problem: a year-end retrospective 2025-12-28 · anthropic.com/news/rag-retrospective retrievalevals