Nice idea having a thread for specific recommendations. The current best options are changing fast so I expect this thread to go out of date within a few months. Maybe even a few weeks.
Here’s my preferences currently (Sept 2024): Sonnet 3.5 - best all-round general model
Opus 3 - best at philosophy and complex theoretical discussions. Usually I only resort to this after I try Sonnet and notice that it fell slightly short at some point in the conversation.
o1-preview (strawberry) - does better on structured multi-step reasoning problems than Sonnet 3.5 Overall, I don’t prefer it, but for something like coding, I’d probably want to at least compare its outputs with Sonnet 3.5′s.
You.com and Perplexity AI - they basically act as wrappers around other models, including Sonnet 3.5 and o1-preview. They add in desirable features like web search, and structured prompt chains for reflective reasoning. They could be handy if you’re into that sort of thing. I personally have tried both of them out, and prefer just doing the prompting and web searching manually over the course of an LLM conversation to solve a problem.
Google AI notebook thingy—haven’t tried it yet, but some people say it’s impressive.
I quite like Exa.ai for search purposes. It’s not there yet to replace my default search, but it’s been useful a few times.
I strongly recommend Cursor for programming. Especially the “apply-model” (where you can press a button to have an AI model figure out how the changes that the chat model suggested to your code should actually apply) makes the experience miles better than vanilla VSCode.
Midjourney is still the best image generator. They now have a website so you no longer need to use their janky Discord UI.
If you’d like to recommend a particular AI product, please reply to this thread.
Nice idea having a thread for specific recommendations. The current best options are changing fast so I expect this thread to go out of date within a few months. Maybe even a few weeks.
Here’s my preferences currently (Sept 2024):
Sonnet 3.5 - best all-round general model
Opus 3 - best at philosophy and complex theoretical discussions. Usually I only resort to this after I try Sonnet and notice that it fell slightly short at some point in the conversation.
o1-preview (strawberry) - does better on structured multi-step reasoning problems than Sonnet 3.5 Overall, I don’t prefer it, but for something like coding, I’d probably want to at least compare its outputs with Sonnet 3.5′s.
You.com and Perplexity AI - they basically act as wrappers around other models, including Sonnet 3.5 and o1-preview. They add in desirable features like web search, and structured prompt chains for reflective reasoning. They could be handy if you’re into that sort of thing. I personally have tried both of them out, and prefer just doing the prompting and web searching manually over the course of an LLM conversation to solve a problem.
Google AI notebook thingy—haven’t tried it yet, but some people say it’s impressive.
I quite like Exa.ai for search purposes. It’s not there yet to replace my default search, but it’s been useful a few times.
I strongly recommend Cursor for programming. Especially the “apply-model” (where you can press a button to have an AI model figure out how the changes that the chat model suggested to your code should actually apply) makes the experience miles better than vanilla VSCode.
Midjourney is still the best image generator. They now have a website so you no longer need to use their janky Discord UI.
Suno is the best AI music generator.
https://www.lesswrong.com/posts/CYYBW8QCMK722GDpz/how-much-i-m-paying-for-ai-productivity-software-and-the