None of what you said contends with the fact that free tiers exist and are advertised (so I assume the providers are serious about them) with specs, and some platforms dramatically outperform others. Whenever anyone gives you a "free sample," they're expecting to convince you that they have a better product. Even drug dealers know this, lolol. Like c'mon.
The free tier is the entry point for most platforms, helping engineers decide on adoption. If free tiers are slow or unreliable, that's an issue with the offering, not the benchmark. Some providers showed strong, consistent free tier results, proving it's possible. A poor-performing free tier is a bad first impression. (... and why should I believe that your more expensive offering is better?)