Whenever they have an alias like this, they usually (always?) have a model with the same checksum but a more descriptive name, e.g. the checksum 38056bbcbb2d corresponds with both of these:
I prefer to use the longer name, so I know which model I'm running. In this particular case, it's confusing that they grouped the qwen and llama fine tunes with R1, because they're not R1.
https://ollama.com/library/deepseek-r1