Gpt4 is focused on state of the art performance.

Claude is focused on state of the art context length and safety while maximizing performance and minimizing token costs.

Sorta like BMW vs Volvo.

Both highly concerned about safety and performance and having amazing features. It's about tradeoffs. Ultimately, I use all available foundational models. Each has specialized into a different niche. All are irreplaceable in my monthly life events.