1. Some users that have free accounts
2. Some of the paying customers
3. Some business customers
4. None of the above and test it with friends etc and release the final product to the customers?
Paying and business customers might not want a beta service that is likely to be buggy. They may want a solid system that is as reliable as possible. On the other hand, they may like the idea of getting new features "first" even if it is going to be a beta. It depends on who your typical customers are: Techies, business types, college students...
Don't give it to the free accounts unless it's already been tested and proven reliable. If someone signs up with a free account to test the waters and then has to deal with a potentially buggy product they might be turned off from your service.
If you have friends who are representative of typical customers (both free, payed, and business) then it would be a good idea to release it to them first. If they are not representative of your typical users you will probably find that you encounter other problems that you weren't aware of during the beta.
If it were my decision, I would alpha test it amongst friends and family until the functional bugs are worked out, then offer it as an option to select (randomized between free, paid, and business) customers to see if they want to sign up for the beta for free (Limited time only!). Make sure they know that the beta is from for x months only. Then when it's time to move it into an add-on product, make it publicly available and offer those who alpha and beta tested it a discount on the service if they want to continue with it.
By the way I looked at your startup and it looks very good. I'm also In Los Angeles. If you ever needed any type of help or just wanted to get together let me know.