There was a time when the EC2 micro instances (t1.micro) were to be avoided like the plague, the CPU was so severely throttled that just setting up the VM for a web server would almost freeze halfway through the process and the response times were extremely high and unpredictable.
Times have changed. With the introduction of the t2.micro and t2.small instances, we're seeing very fast cpus and stable performances over time. How do they do it? Well, modern hardware for one and CPU credits accounting. When the instance is started for the first time, it has more than enough credits to go through the full setup. As long as cpu usage stays below 10% (t2.micro), you keep earning credits, if it's above, you consume them. In the case the web app tested at VpsBenchmarks, the cpu usage varies between 5 and 15% throughout the day so it's perfect. After one week, the instance is at 50 credits (20 above the starting value).
Setup: The setup of an EC2 instance is a bit more complex than Linode, DO or Vultr. First you have to find the type of instance that is right for your needs in a very long list of choices. Then make sure to select an HVM compatible OS for t2.micro and small instances. Setup a Security Group that opens the HTTP and HTTPS ports, create and select SSH keys that work with EC2 (4096 bits is too long). Don't forget that you must login with the 'ubuntu' user on Ubuntu instances. That should be enough to get started.
Performance: Very fast. With an average response time of 36ms, the t2.micro plan is the second fastest provider we've tested. The 1% slowest queries (P99) were served in 180ms, that's pretty decent too but slower than Vultr and Linode 2GB.
Pricing: That's the best part, at $0.013 per hour, the t2.micro instances are cheap. What's the catch? Bandwidth and storage are on top of the "Compute" charges. On our app, bandwidth costs, about 2GB per day, add 50% to the bill. If you're new to AWS, you can get 750 hours of a t2.micro for free.
Stability: No issue to report.