alt Hacker NewsPresumably this has been in production for a while, and is one of the reasons they were able to dramatically lower prices a month ago?
Replies
_0ffh • yesterday at 10:28 AM
Lookahead Sparse Attention should be playing a big role as well, as it dramatically slashes memory consumption.
sourcecodeplz • yesterday at 8:32 PM
good catch, they reduced the prices 75% seems like exactly in line with the speed/inference optimizations gains?
Yes. Section 5 talks about real-world deployment: 5.1: "The DSpark draft models are co-deployed with the preview versions of DeepSeek-V4-Flash and DeepSeek-V4-Pro"; 5.4: "MTP-1 represents the former production setup, having been superseded by DSpark two weeks following the DeepSeek-V4-preview release."