Name and Version

$./llama-server --version
version: 6503 (62c3b64)
built with cc (Debian 14.2.0-19) 14.2.0 for x86_64-linux-gnu

Up to version 6295 (86076f9), the model could continue answering while switching to another message (for example, to read it).

However, with this version (6503), interrupting the model while it is answering will stop the response abruptly. It is no longer possible to switch back to the answering message after leaving it to read older messages.

Operating systems

Linux

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

llama-server

Command line

Problem description & steps to reproduce

Let the model answering while switching to another message (for example, to read it).

First Bad Commit

No response

Relevant log output