I think a lot of people judge these models purely off of what they want to personally use for coding and forget about enterprise use. For white-label chatbots t

I think a lot of people judge these models purely off of what they want to personally use for coding and forget about enterprise use. For white-label chatbots that use completely custom harnesses + tools, Sonnet 4.5 is much easier to work with than GPT-5. And like you, I was really pleased to see this release today. For our usage speed/cost are more important than pure model IQ above some certain threshold. We'll likely switch over to Haiku 4.5 after some testing to confirm it is what it says on the tin.