La Peace

 changes since 1.43

FIXED

- reasoning models like openai/gpt-oss and qwen now get 45 seconds to answer instead of
  12. 1.43 gave them room to think but not the time to do it, so instead of sending back
  nothing they got cut off halfway through. a model on your own pc already got 120
  seconds for the same reason. you can still set aiTimeoutSeconds yourself; 45 is only
  the lowest it will use for a model that thinks first