Today we make a new high performance plan available for the public, Awesome Ape. We've had customers with very high message rate requirements running it for a while.
If you create this plan in AWS it's two c3.xlarge instances with 7GB memory and 4 high performance cores underneath. In GCE it's two n1-highcpu-8 instances, also with 7GB memory but 8 cores(!). In DigitalOcean this plan is two 16GB instances with 8 cores.
Our own benchmarks shows that you can reach about 50.000 messages per second throughput with this plan, with one queue, one publisher and one customer.
Please email us at firstname.lastname@example.org if you have any suggestions or feedback.