Last week brought a day that will go down in history. February 5 saw the release of Anthropic’s Claude Opus 4.6 and OpenAI’s GPT-5.3-Codex. That may not seem especially exciting to the non-expert eye, but behind these arcane names is a new reality.
What makes these new models so important is the fact that much of the programming work behind them was done by artificial intelligence rather than humans. Anthropic insiders are claiming, for instance, that effectively 100 percent of the code behind their products is written by Claude itself. You can debate exactly how much human guidance and oversight goes into that, but other major AI labs all seem to be on the same trajectory. (Disclosure: I am on the Anthropic Economic Advisory Council.)