I've tried the alternatives and even running my own models locally. There is just no other solution I could find that comes close to the general utility of ChatGPT for a broad range of use-cases. I hope someone disrupts them, but until then, ChatGPT is to LLMs what Gmail is to email for me. I love to hate them.
That's an interesting perspective to me since I've gone the opposite way. Claude Sonnet 3.5 beat ChatGPT in every test that mattered to me. Perplexity is a fantastic service when you want gather information or do research a topic, much better than any off-the-shelf LLM imo. I'm not overly surprised that some people have your perspective, since OpenAI has done a great job dominating mind share, but I think if you look at some of the other options you'll find there's a lot of great tools and some of them are even better than ChatGPT. The one caveat I have is that I don't do image generation and it's not very important to me so OpenAI may have the best option there, I don't know.
I have tried other options and you're right, many of them are better at specific things. But I don't have the time or desire to manage lots of tools when ChatGPT is generally good enough for most use-cases.
Claude Sonnet 3.5 beat which version of ChatGPT?
Every single one of them. I’m not being facetious. Sonnet 3.5 outranks every single OpenAI model on most benchmarks.
Claude beats ChatGPT but it lacks many features like audio. Kind of a deal breaker.
What things do you do locally? For me, I found that about 8/10 times - a local model just suffices.
Code and biology. For me, the constraints locally are typically quality and speed of generating responses.