Hacker News

Apertus – Open Foundation Model for Sovereign AI

40 points by T-A 46 minutes ago | 8 comments

maxloh 4 minutes ago

Other fully open LLMs include Allen AI's OLMo 3.1 and MBZUAI's K2 Think V2, both of which have released their full training pipelines and datasets.

Nvidia Nemotron is also an open training source model, though a portion of its dataset remains proprietary.

throwaw12 19 minutes ago

Looks like their instruct models are Llama3.1 fine tune from last year. Is there any progress on new models?

My last hope for soverign AI is from Chinese open models

kordlessagain 8 minutes ago

Sovereign AI is not about using just one model. It's about using the right model for the right job, and getting them to talk through the solution TOGETHER before presenting the answer.

If you want to mix models like this, check out https://github.com/deepbluedynamics/nemesis8

_pdp_ 2 minutes ago

I want to believe.

yreg 31 minutes ago

previous thread: https://news.ycombinator.com/item?id=45108401

trvz 26 minutes ago

The previous version of this model has been pretty bad, but claimed to adhere to copyright laws. However, based on my testing, that's not true either. So in my view this is completely useless.

embedding-shape 24 minutes ago

As long as the following remains true, this release ends up a bigger contribution to science at large than most other models trained "behind closed doors":

> Fully open model: open weights + open data + full training details including all data and training recipes

atemerev 18 minutes ago

I use it extensively. It is not ready for agentic use, but as a generic driving model for RAG use cases, it is pretty competent. You can build useful software with it.