In my experience OpenAI models perform worse on long contexts than Anthropic/Google's, even when using the cheaper ones.
The only reason I open Chat now is because Claude will refuse to answer questions on a variety of topics including for example medication side effects.