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

Is there a "base" version of DeepSeek that just does straight next-token prediction, or does that question not make sense given how it's made?

What is the best available "base" next-token predictor these days?




DeepSeek-V3-Base is the literal answer for what you're looking for (both counts)... but hats off if you actually have the hardware to run it :).


Thank you! I wonder if there's someone out there who is hosting it and providing API access. I've poked around and don't see anything.

Know of a list of available (through an API) "base" models out there?




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

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

Search: