b9967: server: accept null sampling params (#25538)

Release b9967 of llama.cpp, an LLM inference project in C/C++, includes an update to the server component allowing it to accept null sampling parameters (PR #25538). This change enables more flexible API usage by permitting requests without explicitly specifying sampling configuration values. The modification improves backward compatibility and user experience when interacting with the llama.cpp server interface.

Source: llama.cpp Releases | 2026-07-10