Depends on how many elements and samples you want to run, what those elements and their instrumental limits of detection are and how much sample you've got (and the matrix sometimes).
You can already find lots of discussion threads on RG on this, but let's start with this : https://www.perkinelmer.com/lab-solutions/resources/docs/BRO_WorldLeaderAAICPMSICPMS.pdf
Forget about the commercial side of the brochure and read the Instrumental comparison part. This should get you started.
Jamil A. Buledi I would suggest go for ICP-OES. It can detect multiple elements and the excitation is more because of plasma which makes it more feasible.