I was paying openai, but with all the current politics, it looks like I’m probably going to switch to Anthropic.
I don’t have the resources or patience to run my own.
Can someone explain clearly what value you can get from a paid LLM that you can’t get from what is freely accessible?
I don’t pay for them yet, just use the free version for my phone, but Mistral AI seems like and ethical company.
They run a a more lightweight and environmentally friendly model and think they even open-source/weight their models too. I think like 56B parameters instead of like 600b that open ai and antrhopic probably use.
Being a smaller model it’s not as powerful as the big ones, but it’s pretty good!
People pay for this shite? 🤣
none. I have never, and will never, pay for it.
oh the mistake of writing “AI” in lemmy lol. I have made this mistake before, you’ll recover
All of them. Not directly, but RAM and SSD price explosions hurt. Also buying electricity for my home heater from the same market as billion dollar data centers hurt too
eh, i figure we’ll all pay for it some day, when an overzealous capitalist AI harvests the poor for raw material.
I’ve paid for a few months of Midjourney to create some portraits for a few gaming characters. That’s about it.
They’d have to pay ME to use their bullshit, and it wouldn’t be cheap.
I don’t pay for a single one, I’ll use their free services for a specific question to have it summarize the internet basically. I try and avoid chatgpt.
At work we have an internal AI system, and Claude. I utilize both of these to make my workload smoother - but have been finding myself just coding more often than using them because they almost all eventually produce garbage and refuse to listen when you say it’s wrong.
Claude plans for both personal and corporate. My personal is the $100 plan.
And then I have a small amount of credits for API usage on Anthropic, OpenAI, OpenRouter.
Haven’t really jumped on the Cursor train and I think I have Jetbrains AI product sub that I should probably cancel.
Had Github Copilot for a bit a while back but canceled that.
Mind my asking what you do that requires the high tier Claude?
Edit: what you use it for, that is :x
I use the desktop/mobile app for general purpose questions/search/brainstorming.
If that’s all I needed it for I’d be on the $20 plan.
I also use Claude Code for general troubleshooting and some Ansible provisioning assistance on a home network/lab - and some light coding.
Professionally it’s similar usage - but I’m using Claude Code a lot more heavily in writing code and troubleshooting – and am semi-frequently hitting usage limits.
Thank you for actually answering OPs original question and the follow up.
Not much. But I did pay actual money for Sid Meier’s CivilizationⓇ VI, which includes AI that controls the computer players. That’s the one that comes to mind just because its performance was so amazingly weak.
I get the Proton AI with my subscription but barely use it
lumo. it’s (begrudgingly) been useful a few times
I make mine address me as comrade and that’s kinda fun at least.
I keep forgetting Proton has an LLM model available
Perplexity
I run it locally when needed. Can do LLMs and ImageGen, and i’m not stuck with those companies. Installation for LLMs was stupid easy, Imagegen was a bit more involved. Since I’m a Gamer, I already need a beefy GPU, so it’s nice to have a second use for my hardware.
