Episode 62: Codex 0.136, Stanford's Agent Guidelines, AWS OpenAI, and GPU Efficiency