this post was submitted on 29 Mar 2025
822 points (91.6% liked)
Technology
68244 readers
3942 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Have you heard of ollama? You can run deepseek and stuff locally super easy. I know it’s not a complete replacement, but it feels nice to use an LLM guilt free. I’ve compared the 14b distilled model from deepseek vs the paid version of ChatGPT and it made me cancel my account.
What do you use to run it locally? If there was something that could use speech to text reliably to be able to use a open source option, I consider switching.
FWIW speech to text works really well on Apple stuff.
I’m not exactly sure what info you’re looking but: my gaming PC is headless and sits in a closet. I run ollama on that and I connect to it using a client called “ChatBox”. It’s got a gtx 3060 which fits the whole model, so it’s reasonably fast. I’ve tried the 32b model and it does work but slowly.
Honestly, ollama was so easy to setup, if you have any experience with computers I recommend giving it a shot. (Could be a great excuse to get a new gpu 😉)
Yeah, I think the Apple speech to text is pretty decent, but I think on ChatGPT they use the whisper API to return the text and it just seems to be a lot more reliable, especially when it comes to understanding random words in context
How much VRAM do you have on the 3060 to be able to fit the whole thing on the GPU?
I would prefer to run my ais locally, but my brain glazes over if I see github. I found a a program called "gpt4all", but it's very limited in what models it can run, and what I could get just wasn't as good for my use case as openai's 4o model. Also, being able to generate images in the same conversation as text work is a feature that I'm fairly certain no other ai model can do (yet).
I think whats really happening behind the scenes is that the model you’re talking to makes a function call to another model that generates the image.
I haven’t seen it either so if you want that and don’t want to code it might be best to stick with paid, but something like that could easily exist somewhere else.
I bet you're right, but the fact that I never see it is a feature worth paying for, especially for a smooth-brain like myself.