We use a cluster with 2 search peer, one master-cluster-node and one Search head. In our app we have accelerated data models (12), which are pushed to both search peers. The check in the file system show the correct models and via SPL are ok. Furthermore, the search head has the new models too. In the GUI, I am able to see the new data models, but only one data model on the search head is not correct. In the pivot,the eventtypes are missing too.
The datamodels have to be defined on the search-head.
you do not want to deploy them to the indexers, as they will accelerate only their data, and not make it available for other servers (different GUID)