I've been working on an enterprise chat bot for a client for about half a year now. I couldn't help but notice that the parts of their technical team who were adamant about supporting anthropic's models (in addition to or opposition to OAI models) have fallen completely silent.

In February/March it felt like they were ready to cancel the contract over it. Holding my ground was almost impossible because of the intense marketing push that everyone was being exposed to.

Today, I think we all silently agree with the original direction. I can tell that no one wants to die on this hill anymore. I am willing to accept the consequences for the 5.6 model family doing slightly weird things. I've made that clear to the client. I am not willing to do the same for other model families anymore. Not in this context where the client can walk away from me the moment they are unhappy with the performance of the system or the direction it is headed in.

For better or worse, OAI feels like it's become the MSSQL or .NET of the AI world. Annoyingly effective, not (fully) open, sometimes expensive, ran by an "evil" organization, but otherwise wildly predictable. I can actually build a roadmap around this and walk a client through it without them losing track of the rabbit.