alt Hacker News> We match DeepSeek V4 Pro Base using ~50x fewer FLOPs – that’s around half of GPT3’s pretraining compute, or ~$0.5M on GB200.
If this holds up that's a really big deal.
> We match DeepSeek V4 Pro Base using ~50x fewer FLOPs – that’s around half of GPT3’s pretraining compute, or ~$0.5M on GB200.
If this holds up that's a really big deal.
Huge if true. As it were.