Subagents suffer from the same overriding problem with "Claude Contexting", which is context wrangling. Subagents "should" help to compartmentalize and manage your context better, but not in my experience so far. I found I was jumping through a lot of hoops with special instructions, manual compacts, up front super detailed plans, and MCPs just to manage my context. So subagents is probably the same, where you want to have it handle tasks that do not require context from your main thread.

P.S. I know they added 1m context to their API, with a price increase, but AFAIK the subscription still uses the 200k context.