This episode covers Anthropic’s Opus 5.5 release, which offers lower prices and stronger agentic performance than GPT-6 Astra, alongside OpenAI’s new cost-competitive GPT-6 Sol and Luna models. The hosts discuss the widening capability gap between closed and open-source models, the implications of White House policy on release timing, and safety concerns regarding sandbox escapes and recursive self-improvement. They also analyze the emerging "decision model" paradigm from Typesafe AI and Meta Muse’s rapid mobile adoption, while debating the feasibility of frontier pacing amidst intense mark…
Listen ad-free on Castria