I would expect every AI company to use other companies' models in "its own internal tools using special developer access (APIs)", at the they least for evals.

I would expect every AI company to use other companies' models in "its own internal tools using special developer access (APIs)", at the they least for evals. If anything, this is a bad look for Anthropic.