Merged

dhh authored and merged omacom/omarchy#18296 into quattro, adding shared tailnet models beside local models in the agents panel. Users can pair over SSH, start and stop remote models, launch paired models and see when their hosts are offline.

Follow-up quattro commits reduce discovery polling from 20 seconds to two, make pairing and disconnection operate on individual models, and arrange SSH-server setup before sharing. Shared-model usage counts represent machines served within the preceding five minutes. omacom/omarchy@fcb41d664161 omacom/omarchy@ad7bd22a66ae omacom/omarchy@a37ca11dcf05 omacom/omarchy@f83473052679

dhh also merged omacom/omarchy#18515 into quattro to launch Qwen Code against local models and count paired-model usage as local. Qwen launching depends on a companion engine overlay; without it, the launcher falls back to pi. The proposal keeps the model key off command lines.