On Thursday, OpenAI announced GPT-5.3-Codex, a new coding model that’s purportedly 25 percent faster than its predecessor, along with other impressive benchmarks. GPT-5.3-Codex, supposedly, is its first model “that was instrumental in creating itself,” with its team “blown away” by the results. “Holy moly — so it begins!” tweeted another user, who runs an AI newsletter. Taken together, it’s a sign of how the discourse around AI is still dominated by sci-fi concepts and rhetoric and distorted by hype. More on AI: Investors Concerned AI Bubble Is Finally Popping