If that's the case, do Anthropic have an even better one helping them? The Chinese labs too? Where this OpenAI "advantage" is taking them?
No? The top of human engineers are much better than any model would be, so AI models really aren't a big advantage when you're trying to develop anything that is SOTA.
Not every problem is best addressed by a top engineer.
Plenty of the tasks that keep a company running can benefit from good-enough (and better than the competition).
Yes, and none of those tasks require even the current SOTA models.
You can't fathom how 'top human engineers' could take advantage of exclusive access to a frontier model at ultrafast speeds and limitless token budgets to conduct research?
I have all that and LLMs invariably produce garbage, so, no.