From an Ars Technica article[0]:
"But far from getting cut off suddenly, Parler had months of warning, Amazon says. Amazon's filing included copies of emails it sent to Parler in mid-November (PDF, content warning for racial slurs) containing screenshots full of racist invective about Democrats, including former First Lady Michelle Obama, with a series of responses from other users to "kill 'em all."
And
"Representatives from AWS spoke with Parler executive leadership on both January 8 and 9 about the platform's "content moderation policies, processes, and tools," Amazon said. In response, Parler allegedly offered steps that would rely on "volunteer" moderation, and Parler CEO John Matze allegedly told AWS that "Parler had a backlog of 26,000 reports of content that violated its community standards and remained on its service."
So not so much.
[0] https://arstechnica.com/tech-policy/2021/01/filing-amazon-wa...
The better comparison would be instances of hate speech or calls to violence that were directly reported to Twitter by its hosting provider for the express purpose of requesting that they be taken down. I'd certainly give you even odds that the number of such reports that Twitter both received and ignored during that same time frame was fewer than 98, despite the fact that Twitter is orders of magnitude larger.
No doubt there are instances of similar behavior that go under the radar on Twitter and every other comparable site, but once your hosting provider has contacted you about a specific instance that's probably a sign you should do something about it.