Azure Analysis Services offers three tiers – Developer, Basic and Standard.
Developer Tier is like developer edition of SQL Server Analysis Services i.e. includes every feature supported by AAS but with a difference that you can use it for your production workloads.
Basic Tier is like Standard edition of SQL Server Analysis Services. It doesn’t support perspectives, partitions and DirectQuery.
Standard Tier is like Enterprise edition of SQL Server Analysis Services and supports all the features.
Each tier has different performance levels that drives the costing of running AAS in your environment.
Developer tier has only one performance level that offers 20 QPUs and 3GB or Memory
Basic Tier offers two performance levels B1 and B2
Standard Tier ranges from So to S9
You can change the performance level within a tier i.e. you can go anytime S0 to S4 and vice-versa and upgrade the tier too but downgrading the tier is not permitted i.e. if you have upgraded to standard tier, you can’t switch back to basic tier. One trick you can try to downgrade the tier is to take the backup of your cubes, setup a new instance with the new tier you are looking for and then restore the cube on this, remove the old instance.
Performance of AAS is measure in QPUs, Query Processing Units. QPU is a blended measure of vCores + Memory. Microsofty says to assume 25QPUs as roughly 1vCore so if you are trying to map your on-premise server to AAS to determine how many QPUs would be equivalent to your on-prem cores, you can consider to multiply the cores by 25 and those many QPUs, you would need. However, the primary advantage of cloud is to always start with little less than what you need an then upgrade.
As of writing this post, it is how it looks like –