Comment by nkko

Yes, this is frustrating, but it doesn’t occur in CC. I run the conversation logs through an agent and opencode source, and it identified an issue in the reasoning implementation of opencode for Zai models. Consequently, I ceased my research and opted to use CC instead.