Anthropic released a model being called Opus 4.8, but the headline is misleading. This is not a true Opus 4 successor. The video breaks down benchmark comparisons against GPT-5.5 and argues that the distinction between model updates and platform updates is being blurred by both Anthropic and OpenAI, a pattern that matters for anyone making infrastructure decisions.
OpenAI's Codex is getting serious platform expansion: Windows computer use, a mobile app, browser upgrades described as significant, and the ability to spin up nested Codex chats. The section on Codex browser capabilities runs from 07:40 to 10:38 and is the most technically dense part of the video. There is also coverage of developer migration away from Replit and Lovable, which signals a real shift in where builders are choosing to work.
The video is worth watching in full for the agent mini apps section starting at 25:42 and the unfiltered closing commentary. The host is building Chorus, a tool that embeds Codex and Claude Code into iMessage, so the perspective here is practitioner-level, not analyst-level. The Replit and Lovable exodus discussion alone justifies the runtime.
[WATCH ON YOUTUBE →]