LangGraph v1.1.3 ships deep-agent templates and a distributed runtime.

Per Requesty, LangGraph now exposes prebuilt planner/researcher "deep agent" templates and distributed runtime support through the CLI, on top of earlier state-persistence and human-in-the-loop checkpoint work. The distributed runtime targets long-horizon multi-agent graphs that outgrow a single process, which is a real wall you hit fast once your agents start spawning agents. If you're already on LangGraph, the templates cut boilerplate. If you're not, this is the kind of feature that makes the framework hard to leave.