|
|
|
|
|
|
by user43928
11 days ago
|
|
|
They are not good enough. Anecdotally, I was wrangling GPT 5.5 at work today, trying to get it to implement tests in the same style as a reference project. I could hardly believe just how dumb GPT 5.5 Medium acted. It took around five turns to iron out the obvious errors and idiotic inventions. I am curious to evaluate the same use case with Fable or 5.6 Sol, but my 'AI-first' employer only offers access to outdated models at a laughable budget. |
|
|