logoalt Hacker News

crostlybostly • yesterday at 5:05 PM • 1 reply • view on HN

Is AMD inference clunky these days or does it work pretty seamlessly with most local tools?


Replies

trollbridge • yesterday at 8:16 PM

Works fine if you want to run Qwen, etc. or do OCR stuff.

I really only have it for data locality - some customers simply don’t have approval for me to share data from their customers with random cloud inference providers. I also use Bedrock for that but it’s expensive