Using Claude regularly again has made obvious one of the biggest differences between Anthropic and OpenAI models: taste
GPT (Astra included) just doesn't really compete. Sure, you can get ~similar quality outputs with enough prompting, time and effort, but Opus 5.5 just combines great taste, intuition and vision in a way no model before it has. While being incredibly usage-light in Claude subs, something that cannot be said about Astra
A big step in the right direction in all sorts of ways for Anthropic, and I'm excited to see what they have in store soon for Haiku, Sonnet and Fable