Few would disagree with "better" (than either pure threads or pure events) but if you're going to say "best" the burden of proof is higher. Please explain why you think actors are strictly better than e.g. promises/futures or agent-based models.
Keep in mind that actors and futures can be used together, which produces an even more powerful result than either abstraction used alone. We use this programming model in Mesos via a library called libprocess. There will be some talks on this approach in upcoming MesosCons and we'll submit one to Cppcon.