If you mostly use chat-style AI tools and are only starting to track new infrastructure, this is the part of Iroh 1.0 that matters. The mistake is to read it like a normal feature roundup. If you do that, you can spend time learning the wrong model and miss the only decision that actually changes your next step.
Here is the useful verdict: if you are building with Iroh, the first thing to delete is host:port. That sounds small, but it is the whole shift. A product update is worth your time only if it changes your next decision, not if it just adds more features. This one changes the starting question.
The June 15, 2026 1.0 launch said it plainly: Dial Keys, not IPs. That line matters because it moves the address away from a changing IP and toward the device key. The README pushes the same idea from the developer side: you point at the target device, and Iroh looks for and maintains the fastest connection.
The docs make this concrete with EndpointID. In plain English, each node gets a stable identity, then Iroh handles the messy network work around it: address discovery, NAT traversal, and relay fallback. You do not need to memorize the jargon to get the point. The old question was which host and port do I dial. The new question is which device am I trying to reach.
That is why I would not file Iroh 1.0 under faster P2P. The more important change is the mental model. If you are evaluating it, check your own plan first: are you still designing around host:port everywhere, or around device identity first? If the answer is host:port, that is the part to rethink before anything else.
Keep this as a short filter, not a full market verdict. This read is based on Iroh 1.0 launch and docs language as of June 15, 2026, not on community feedback or competitor comparison. Share it with anyone who is about to learn Iroh from the feature list instead of from the address model.