Agree. The one thing that kept pushing me into a second session was wanting another model's take. So we made it possible to add that model to the same thread, with a short catch-up on what it missed.
Depends on the job. Opus 5.5 does most of our building and either GPT-6.1 Sol or GPT-6 Astra reviews it. Having one check the other has been more useful than picking a winner.
Same combo here. Opus 5.5 builds, either GPT-6.1 Sol or GPT-6 Astra reviews before anything ships. This morning the review caught six lines in our release notes the code didn't back up.
Yeah each thread always has a main agent who owns the room and the added agents only work based on what the main agent tells it what to do or you as the user
Threadlines v0.5.0 is out 🧵
Rooms: more than one coding agent in a thread. Ask Codex while Claude works, or have them review each other.
Also new: Cursor and fx support.
Free and open source.
Release notes: github.com/Threadlines/threa…
100% Agree, we build out newest feature around that. We call it “Rooms”, you can have Agent work together in the same thread, been using it in nightly for a bit now and I love it. threadlines.dev/