- Dax says he usually doesn’t hype new models. But GPT-5.6 has noticeably changed how his team works.
- The key point: he doesn’t think the model writes good code on its own. What he likes is not the out-of-the-box result, but the process of getting to good code with it.
- According to him, GPT-5.6 understands the task better and loses fewer details along the way. It also catches things the developer may have forgotten.
- The problems are still there: the model can still be overly defensive and write unnecessary code. Sometimes they do another pass afterward and delete half of what it wrote.
- The most telling fact: his team’s token usage is up 5x. That may be the clearest signal that the model has become a more useful coding assistant, not a perfect code writer.
