OpenAI GPT‑5.3‑Codex and Anthropic Claude Opus 4.6 announced

Capable, very credible AI models are being delivered at an increasing pace, and earlier this week, both the big names, OpenAI and Anthropic, launched their newest models almost simultaneously: ChatGPT-5.3-Codex on February 5, and Claude Opus 4.6 on February 6.

Both products are very capable of complex software development, and the benchmarks are stunning. I had time to test Claude Opus 4.6 briefly, and the results are impressive. What might take a skilled developers a week or more, can be completed by these new AI models in just hours. I have tested both simple needs, but also more complex requirements. To these models, my short tests weren't a problem at all.

All this leads to the important question: if generating code is no longer the bottleneck, the question becomes: what should we build?