DeepSeek certainly managed that on the training side but in terms of inference, ... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		thom 15 days ago \| parent \| context \| favorite \| on: OpenAI's cash burn will be one of the big bubble q... DeepSeek certainly managed that on the training side but in terms of inference, the actual product was unusably slow and unreliable at launch and for several months after. I have not bothered revisiting it.

snuxoll 15 days ago [–]

Are you talking about the model or their service? There's plenty of options for using their models other than the official DeepSeek API.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact