Actor model (Akka/Orleans) for high-concurrency systems: still relevant?

I'm curious about the current relevance and practical applications of the Actor model (like Akka or Orleans) for building high-concurrency, distributed systems. Historically, it's been powerful for managing state and concurrency without explicit locks. Is the Actor model still a go-to choice with the rise of containers, Kubernetes, and serverless architectures? Or do these new primitives offer simpler ways to achieve similar scalability and fault tolerance? I'm also looking for operational experience reports, especially around actor persistence, recovery from failures, and how it integrates with modern observability stacks. What are its sweet spots and where does it introduce unnecessary complexity?