aragornii

I've been using both Opus 5.5 (high) and GPT 5.6 Sol (medium) intensively for the last few days and I'm noticing more and more that, even though Opus gets most of the things correct (and it's great when it comes to visuals), GPT Sol is the only one able to identify edge cases and inconsistencies.

I wouldn't say Opus has been nerfed. They are just two different models but, at the end of the day, GPT Sol is the model I trust the most, at least for now.

I'm trying to test them but the Visual Studio Codex Extension on WSL2 is not helping.

It seems there's a bug, shipped together with the flag that enables the new models, that doesn't allow Codex to run properly in the WSL2 sandbox.

What I'm mostly interest in is the Communication section. Opus 5 was so convoluted in the way of answering that was really frustrating me.

Instead of instilling confidence, it was overwhelming. Not sure if I'm the only one.