Loading…
5
Spent $200 on a 3060 with 12GB to run local models and I keep going back to the free cloud ones
Half the people say local AI is the future of privacy and the other half say cloud models will always be smarter, so which is it? I dropped $200 in March on the card plus another $80 on a bigger power supply, and my little setup runs a 7 billion model fine, but it still gets stuff wrong that the free stuff nails on the first try.
1 comments
Log in to join the discussion
Log In1 Comment
nancy_davis24d ago
Wait, 200 bucks for a 3060 with 12GB? That price seems too good, I thought those were going for way more even used. But man, the part that gets me is you spent another 80 on a power supply on top of it and you're still not happy with the results. That's almost 300 dollars to basically end up using the free stuff anyway, which has to sting. I've messed with a 7B model before and yeah, it'll write you a decent email but ask it anything with real facts and it just makes stuff up with full confidence. Meanwhile the free cloud thing answers the same question right the first time and I'm sitting there wondering why I bothered.
10