My Opus 4.5 Moment

The "ChatGPT moment" was November 30, 2022, the day OpenAI released the first version of ChatGPT. It wasn't just a technological leap; it was a usability leap that took an LLM and put it behind a simple, free chat interface that reached 100 million users in just two months.

If you had asked me a year ago about LLMs for coding, I would have explained they were helpful but overhyped. They wrote some functions I would refactor, and I often got stuck in a loop of pasting error messages with "fix this". But then, just 3 years after the first ChatGPT, on 30 November 2025, Anthropic released Claude Opus 4.5. I realized I was wrong: I had my Opus moment.

Opus 4.5 is shockingly good at coding (as are the other massive models released end of 2025; Gemini 3 and OpenAI Codex 5.2). It nails almost everything I throw at it. I'm now at the point that I hardly program myself anymore, and for hobby projects I don't even bother reading the code anymore. To make the point: I even built a macOS app (safekeylogger) in Swift, a programming language I have never used. Heck, I have already finished more side projects than the last 5 years combined.

It's a transformation. Software engineering is changing rapidly. It brings mixed feelings. I've always enjoyed programming because I love building and getting things done. I have spent countless hours mastering programming skills, something I need to do so much less of now. However, the main feeling is excitement for what to build next.

I'm not the only one to have had that Opus moment. In July 2025, David Heinemeier Hansson (influential programmer) argued he still prefers to "chisel code out of thin air with his bare hands in an editor" (paraphrased, from his podcast with Lex Fridman). But now he's ready to "give the LLMs a promotion" saying they are "fully capable of producing production-grade contributions" and "you don't have to connect many future dots on the current trend line to get dizzy by the prospects" (Promoting AI agents).

Gergely Orosz is another influential voice (known from his books as "The Pragmatic Programmer") who wrote in early Jan 2026 "I’m coming to terms with the high probability that AI will write most of my code which I ship to prod, going forward." (The grief when AI writes most of the code).

Peter Steinberger (a well-known iOS developer) puts it like this: "Whereas in ~May [2025] I was amazed that some prompts produced code that worked out of the box, this is now my expectation." He states "I can ship code now at a speed that seems unreal... the amount of software I can create is now mostly limited by inference time and hard thinking... and most software does not require hard thinking" (Shipping at Inference-Speed). He also mentions the huge leap in usability with the late-2025 models: "The step from GPT 5/5.1 to 5.2 was massive."

Others are still hesitant. Vicky Boykis (a well-known ML engineer) is worried about "being overrun by mediocrity and sloppiness". She has a point: the best code is still no code, and great programmers will always recognize 'the claw of the lion'; the mark of mastery (I want to see the claw).

I believe we need to start building. As Wes McKinney (creator of pandas and arrow) put it: "Too expensive, too much time, not worth it, too little marginal benefit, does not benefit enough people. That's over now." Instead, he argues we should ask Why Not?, as we can build software for one person. Heck, we can build it for one-time use.

In the last couple of months, I've built many things. For example: an extensive site for comparing alternative keyboard layouts (alt-alpha), a three.js visualization of star constellations in the midnight sky (this has been fun to learn!), and even one-shot a rapid serial visual presentation webpage to read text really fast. I also created an overview website of some of the tools I've built at tools.timvink.nl, with basic prompts like create a static html tool overview website and Add app X to the overview. There is so much more that's possible.

A couple of closing thoughts: the bottleneck now is agency, the creative part of deciding what to build. Here experience counts; demand for senior engineers will go up, not down. If you build more highways, it becomes more attractive to use the car, and traffic jams can actually increase. The cost of software engineering is rapidly going down, which means it is now feasible to build so much more. We're only getting started on what computers can do for us... the future is truly exciting!