Episode notes

Six mystery models showed up, dominated the coding charts, then vanished—now the AI world thinks we just watched GPT‑5 get built live. Yes, live. And OpenAI might be using these fragments all the way through GPT‑8.

We’ll talk about:

  • Why models like Summit and Zenith might be GPT‑5 pieces in disguise
  • How OpenAI is using real users to secretly test “mixture-of-experts” systems
  • What makes Claude and Grok look slow compared to these
  • China’s bold new AI plan and why it’s challenging U.S. dominance

Keywords: GPT‑5, Zenith, Summit, Claude Code, LM Arena, MoE, AI Action Plan, China AI, Claude Sonnet, OpenAI, Grok 4, Copilot

Links:

  1. Newsletter:
 ...  Read more
Keywords
OpenAIClaude CodeChina AIgrok 4GPT‑5LM ArenaSummitZenithCopilotClaude SonnetAI Action Plan