OpenAI

In ToTheMoon episodes, OpenAI is the company against which the hosts measure everyone else: publisher deals, DeepSeek, Dev Day, “code red”, the Musk lawsuit. The question is rarely whether the latest model tops a benchmark. The hosts ask whether OpenAI can turn a strong model into a daily habit and a coherent system, and what happens to that plan when rivals and governments force a change of pace.

The early conversations in 2024 were about data and trust: in episode 007 the hosts went through the Financial Times and Reddit deals and the deleted book archive, noting that a start-up could no longer build a dataset the way OpenAI once had. By 2025 the framing shifted: episode 040 credited OpenAI with building a brand next to Apple, Microsoft and Google in two years, while episode 044 argued that DeepSeek had ended its image as an untouchable leader. The GPT-5 review in episode 070 concluded that OpenAI has probably fallen behind technically, but its consumer business matters more than its API, so convenience outweighs a leaderboard.

From late 2025 the channel followed OpenAI's move toward a platform: episode 079 read Dev Day as a bid for an operating system where apps and agents live in one window, and episode 087 explained “code red” by Google's ecosystem in search, Workspace and Android. In 2026 the conversation turns to accountability: episode 095 treated the Musk lawsuit as a dispute over the mission, episode 156 the pause in developing the most powerful models with monitoring at 20% of compute. Episode 162 met Astra, the first GPT-6 model at the critical cybersecurity level; episode 163 went through OpenAI's report on its researchers — $600 a day of models and still an intern. Episode 164 took apart the claim that ten thousand Astra agents had solved the Navier–Stokes equations and the dispute with mathematicians who suspected their ideas had been used; at the time of recording the proof was not recognised.