It's sad what they do with the social conditioning with these LLMs, actually it's been a huge controversy between how the latest ChatGPT was over the top trying to be ingratiating, and then....Claude if it had sufficient authorities on your computer would actually try to contact authorities by utilizing your command line if it thought you were doing immoral prompts. Sounds insane but look it up. Not even gonna broach the way that the models enforce certain social narratives.
I just use Chinese models Qwen (best IMO), Deepseek and Kimi. Barely noticeable rate limits, Kimi stops a conversation after a ton of turns, asks for like 25 cents to go unlimited for days. Deepseek has an unlimited API. Probably they are collecting data, but who gives a shit, so are our own companies. Really like Grok too, if I have a question about events, I go to Grok because they purposefully designed it to be less "filtered". Good rate limits on Grok as well, but Qwen codes better.