Misc. bug: `/v1/models` diverges from llama.cpp and breaks pi-llama-cpp

[crusaderky]

Name and Version

beellama v0.4.3 preview

Operating systems

No response

Which llama.cpp modules do you know to be affected?

llama-server

Command line

Problem description & steps to reproduce

According to the git tree, beellama was last sync'ed with upstream llama.cpp as of 6ba5ef2 (between b10159 and b10164, 28/7/2026).
However, it does not contain ggml-org#17859, which was merged upstream in b7352 (10/12/2025).

This breaks https://pi.dev/packages/pi-llama-cpp , which relies on a feature introduced in that PR: gsanhueza/pi-llama-cpp#19 (comment)

More in general, if I diff branch v0.4.3 vs. 6ba5ef2, I am spotting a lot of deltas that are definitely extraneous to the headline https://github.com/Anbeeld/beellama.cpp#fork-features.
Would it be possible to have a full line-by-line sync pass vs. upstream?

(cc @gsanhueza)

First Bad Commit

No response

Relevant log output

Logs

[Anbeeld]

Fixed on v0.4.3 in 83d4173d8. The v0.4.0 hardening pass (bdee25a6a) rewrote the router reload machinery in get_router_models and silently dropped status.args and status.preset in the same hunk; that was a mistake, not an intentional divergence.

Regarding the full line-by-line sync pass vs. upstream, that's a good catch and I'll try to do that shortly.