Our Minimum Viable Product will be an online service. Development takes some time.
Meanwhile we would like to release a small but supposedly impressive feature for free. The goal is to get some feedback and promote ourselves.
Which one of these 3 tactics would you recommend?
This is quite a complex scenario. Your stage in customer development isn't clear to me - I wonder, how can you have an MVP if you've done no customer development?
If you still have some major assumptions to test about the customer segments you're targeting and the problem you're solving for them, your focus is "customer discovery." That is, your immediate goal should be to get that feedback. Trying to build any MVP at this stage is premature and puts all your effort at significant risk of not having market traction.
If you've reached the point where you and potential customers have isolated the problem and they have described ways they've tried, and failed, to solve this problem themselves, then you can start to look at validating your MVP.
I would test that small feature set as if it was your MVP. After all, your MVP will change as you learn. This will also give you base from which to test other MVPs.
Whether or not to charge is a separate issue, and your time will be better spent dealing with that when you are validating your MVP.
Hope this helps.
I'd vote for releasing the small feature for free and getting some feedback as early in the process as possible.
What about quick-and-dirty screenshots and "smoke and mirror" videos?
Depending on the application this could be a better way to express your ideas to potential customers and to gather feedback. While positive responses aren't really validation, you might get a lot of information of there's an unexpected level of disinterest.
Base it on how you came to the conclusion this service will be impressive. As a typical customer yourself, you may have been sold on the idea itself. It helps if you are knowledgable on the subject and can conceptualize the idea.
Did other people "get it" by hearing your story? Did you have to diagram it on a napkin or whiteboard? You could get away with a mockup.
If it didn't come together until someone wrote some code, you should think about the free release.
The second criteria is how long until you release the product itself? If you can get the free version going soon, then wait. Otherwise, you may want to do a mock up.