On Tue, 22 Sep 2026 19:34:33 +0200,
Guillaume MULLER <guillaume.muller@emse.fr> wrote:
> 
> Dear all,
> 
> I'm using llama.cpp for some projects. The version that is in the packages (for OpenBSD 7.9) is very old and cannot run most of the current models.
> 
> So I downloaded and compiled the latest version from the official GitHub repository. It works perfectly fine without any modification (with just CPU support).
> 
> I wanted to ask the maintainer to update the version, and I just noticed that it is in fact already in the ports!
> 
> Would it be possible to expose it as a package, instead or requiring manual compilation or installation of the whole ports tree?
> 
> Thanks you very much.
> 

I keep llama.cpp and libggml synced with latest releases and usually an
update follow in days after new release happens.

I not sure that backport llama.cpp and libggml is right to -stable, and I
suggest you to use -current instead.

-- 
wbr, Kirill