Loading…
14
Small local LLM beat my API budget for customer tags
I was burning $220 a month on API calls to classify support tickets, so I tried running a 7B model on a used GPU I got for $300. It nailed 94% of the same tags after I fed it 500 old ticket examples, way better than I expected. Anyone else switch from cloud AI to local models for a boring task like this and see real savings?
1 comments
Log in to join the discussion
Log In1 Comment
felix1557d ago
How long did it take to hit that 94%?
0