After a few days of Claude 3 Opus I am sold on it vs. the ChatGPT or OpenAI API’s. It’s like night and day, ChatGPT can’t code that great vs. Claude when you start building huge Rust based projects. Claude really can take 2 files and merge them vs. ChatGPT just screws that one up every time. It’s like ChatGPT is gaslighting you into believing less output is better, Claude just delivers. Sure Claude has flukes here and there but each one is not a show stopper plus using ChatGPT for initial templating can be a good combo (since it’s not as rate limited for now). Yet when Claude is not rate limited like now, why would I use ChatGPT again? I mean just code with Claude and see yourself, it’s made me feel these “Lazy” discussions have jumped the shark when we realize the model is just outdated already.
I feel OpenAI benefits from us saying “it’s limited by guard rails” vs. saying maybe the model is flawed and out of date now, we have seen too much evidence htat points to that potential reality already. Claude shows us that there is a higher level quicker achieved than OpenAI, who claims they are holding back when perhaps that is just a tactic to look cooler to people investing their AI in OpenAI’s models. Promising what can be done by Claude yet “we are being careful” is a great air guitar / air ware future carrot on the stick tactic.