Been happening already. Not for api token prices yet (although they are increasingly locking new models behind request forms on aws and azure), but on the public-facing web apps they are pulling everything. Claude you used to basically be able to talk to sonnet 4.6 all day long before it cut you off now it gives you like 3 messages. And codex I used to be able to write like 10K lines of code in a week before it cut me off. Gemini on AI studio is more limited now too. They banned one of my burner accounts last week too lol.
I'm on the basic plan. it might be the same token amounts actually but is just faster now idk. i just hit the 5 hour limit the other night for the first time though. it was a really aggressive loop though and it is way more accurate than it used to be and I was using it for some simulations I was running locally and didn't have to push it to github or move between containers or anything so I might have just been burning through tokens faster than I used to.