Try lowering thinking level with GLM-5.1, to me that seems to have an impact on mitigating the blocking behaviour.
replyHmm I'll try that, but OpenCode shows me the thinking and it's not even doing that. I'm just getting no tokens from it at all.
reply