The physical work? No idea.
From a risk and compensation standpoint, probably much better at Amazon. If you crash your car driving for Uber Eats, you're on the hook for it. If you're injured on the job, you simply don't earn money since you're an independent contractor.
Not to mention health, life, and disability insurance and PTO all provided by most large companies, and having a point of contact who might care about you on a personal level as opposed to getting all your work assignments from an app optimizing profit not for you, but for the service company.
My employer ships coolers of food to people. The fulfillment centers are cold, and it's really hard work. But if John's car breaks down, the team works together and figures it out. Mary agrees to swing by and pick John up on her way to work, and Mark will take him home after the shift until the car is fixed. The manager sometimes picks people up herself.
Of course the workers will be exploited anyways because lower costs means more profits but they don't need to be exploited quite as much. At the end of the day the workers will be exploited exactly up to the limit of what the company can get away with. For "disposable" and interchangeable untrained workers like delivery people that limit is pretty high.
This argument only works if you consider 15-minute delivery as a different business unit as 2-day delivery, so they'll be fighting it out to gain market share. In reality, the same company is behind both options, so it doesn't really matter. They really don't care if you choose 15-minute or 2-day delivery, only that you bought it from them. If anything, they'll keep the 2-day delivery cost the same and charge a premium for 15-minute, because "how urgently you want it" is a pretty good price discriminator.
does it? 15 minute delivery definitely costs more so you could argue that margin is lower, but at the same time you could argue that 15 minute delivery is a premium product that can command a higher margin.