I've been really frustrated with 4o lately. It feels like it's constantly contradicting itself, remembers only part of our conversations, and overall just isn't as reliable as it used to be. I really enjoy using o3, but I'm aware it's not very versatile. Most of my use for these AIs is for research and helping me write or respond to emails. I do have an iPhone where it's integrated, but honestly, Apple's AI isn't much better. I've been hearing a lot about the new Gemini paid model and rumors of Apple potentially partnering with Google for integration instead of ChatGPT. I'm curious if it's worth looking into, and I'd love to hear about everyone's experiences! Thanks in advance!
5 Answers
If you're into AI, consider building a beefy gaming rig to run a local distilled model. It lets you keep chatting regardless of internet issues. Also, check out Google’s AI studio and try fine-tuning a model to your liking.
Try using o4-mini without the high settings. You get a limit of 300 messages daily. Don't let people scare you off with claims of it being hallucinating; it’s worth a shot!
Honestly, I think Gemini is a solid alternative right now. Just be aware that things change all the time, so check back for updates later! It's wild how fast everything is evolving.
If you're running into contradictions, remember to clear 4o's memory regularly. It's been improving, but you really need to keep things clean. Otherwise, it can get off track quickly.
Aistudio from Google gives free access to all their models, but I found it lacking in some comforts that ChatGPT offers. Plus, 4o seems to have reduced its output length significantly—where I used to get around 1450 tokens, now I'm lucky to hit 800. It's frustrating because the answers feel way too restricted now.
I took your advice and cleared its memory! I also turned off the chat history option since I like to refer back to my old chats. I added important info to my custom instructions instead. I’ll check back in a couple of days to share if I notice any improvements!