Having Pro accounts with OpenAI and Claude, I traded GPT4 to Claude3.
Most of my tasks are in software development.
Cannot say GPT4 is dumber than it was, but Claude3 solves my tasks significantly more accurate. Better comprehension of my specs.
And it is equally applicable to:
- simple tasks, like writing a simple batch/bash script,
- or medium level tasks, like generate SQL query, based on specs,
- or complex tasks, like generate C# complex logic source code that does complex database manipulation.
It’s sad to walk away, but these are my realities.
Any comments?
Maybe I am missing some magic button in GPT4 settings (I love magic)?