Multivariate testing is more advanced than A/B testing. In a multivariate test you are testing multiple variables in an email or landing page and how they interact with one another. The goal is to determine which combination of variables performs the best out of all of the possible combinations.
For example, for an email you could test two subject lines and two calls-to-action to see which combination resulted in the best click through rate. This test would give you four variants of the email (2 subject lines x 2 calls-to-action).
For multivariate testing of emails you send each variant to a percentage of your contacts. Say your sample size is 100,000 people, you can send each variant to 15,000 people (15% each). A winner is determined based on your desired metric (most opens or clicks) and then that winning version is sent to the remaining 40,000 people (40%).
Like A/B testing you should always have a goal in mind when conducting multivariate testing. Figure out the problem you want to solve and identify the variables that need testing. In addition, to perform multivariate testing you need a much larger sample size than for A/B testing to obtain meaningful results.