Our builds machines actually have an 8gb limit for memory, regardless of the instance type, so you shouldn’t need to switch to a different instance type if your app just requires 1 - 1.5 GB of ram during the build.
I hope this helps - let me know if you have any further questions!
Taking a look at the failed deploys, it looks like you are hitting the maximum allocation heap memory on the service startup, which is not during the build phase, but after it. This will be subject to the instance type limits specified here.
I also looked at some of the historical metrics for your service, and do see that it has bumped into the memory limits for Starter in the past.
At the moment, we do not have the option to select higher limits for the service startup vs. when it’s running after, so you would need to select the instance type appropriate for your app’s resource usage on startup as well.