#Internal Reason visible on Telegram

1 messages · Page 1 of 1 (latest)

rough trail
#

Hi,
I don’t want to see the internal reasoning of Qwen3:30b which I am using as the local model. The reading and thinking should be switched off, which was confirmed by the dashboard and I tried this via the telegram command as well as in the OpenClaw config json. Still, the chatter continues. Is this related to the model, which prevents openclaw to not see the think or thought tags? How can resolve this or dig deeper?