ORTC is basically the same as WebRTC nowadays (it was more different back in the day).
The gist of what you need to know is that many WebRTC tools and browser’s RTCPeerConnection work with SDP (legacy text-based format), mediasoup uses typed object-centric data structures (mostly introduced in ORTC) instead. So you will not see SDP on the wire or in API, but you can bridge tools using SDP with mediasoup.
It is not either or, mediasoup works fine with browsers and multitude of tools (especially when using plain RTP that mediasoup also supports).
Does that mean the mediasoup-client uses the browser’s RTCPeerConnection and then parses the SDP messages to objects? Or even in the browser it manages to avoid the SDP messages?