Ask HN: Which AI subscription is worth paying for now?

Posted by 0x70run 7 hours ago

Alternate title: Which AI subscription are you paying for right now, and why?

Currently I'm not really paying for any AI subscription. I'm largely reliant on the company's Claude offering but I'd like to separate my personal stuff from work. I've checked out the following ~20$/month plans and have some thoughts:

* Claude seems to be the most capable, but the app was pretty terrible when I tried it a couple of months ago, I rant into frequent rate limits, crashes, etc. It was just not that reliable overall.

* ChatGPT has the best app overall (project folders + end to end experience was super nice), but I'm not sure about the ads and if the company's going to be around after a while...

* Gemini is OK. The app is pretty bad again, no support for folders/projects + I'm trying to de-Google my life a bit so I'm not sure if I should go all in with this...

My usecase(s): I would like to build smaller applications/scripts on the side to fix annoyances in my life and automate things every now and then. Every now and then there might be a hobby project. I also intend to use LLMs for slightly more involved rubber-ducking to work on ideas that span across programming, music (production, performance, learning), and literary fiction.

Comments

Comment by resonious 7 hours ago

The subscriptions are per month so I wouldn't use "will the company last" as a deciding factor. On the contrary, if OpenAI really is on its way out then I'd say get their subscription and use the heck out of it while you still can.

Maybe this shows already but my answer would be to use OpenAI just because I think codex is better than the others.

Comment by twosdai 7 hours ago

You could consider making your hobby project to get a local AI stack up and working.

Open source models are not as good as claude, or GPT. But they don't need to be to fulfill useful small tasks. I think its worth considering using claude from your work to help you setup a local harness and model and then use that.

To me it feels a bit more fulfilling.

Comment by 0x70run 7 hours ago

I would love this tbh, but it's just that context window issue for me right now. For my personal use, I have a MacBook Air M5 (maxed out), but I don't think I can run anything decent with large context window (for longer chat sessions) on it. I should also mention that I'm still learning about LLMs, so if I've missed something please do let me know. Running local LLMs is something I'm very interested in.