In my experience, Gemini 3.7 is excellent for general non-coding tasks. But for coding, especially backend development, I still find models like Opus 5 and GPT-5.6 more reliable.

If I strip the dependencies out, I'm using it on a 5MLoC C++ codebase, and I found it performs really really well. I am using the Opus 5/Fable in parallel and I couldn't tell the difference. Both models make mistakes here and there.

I’ve been using Gemini for coding daily since 3.5, mostly on some pretty complex ML engineering projects. It’s great.

What’s an example of not being “reliable”?

The default mode already does something along those lines — certain elements take two hits to break. But for the hidden B-side mode I went in the completely opposite direction: the ball phases through everything by default. See if you can find it ;)

Calling them 'laws' is always a bit of a stretch. They are more like useful heuristics. The real engineering part is knowing exactly when to break them.

Especially since a lot of them are written in a tongue-in-cheek way. And many are contradictory.