Async-RBench Evaluates Asynchronous Task Execution in LLM Agents
October 26, 2026
Async-RBench introduces a benchmark of 200 tasks to evaluate how LLM agents manage asynchronous subagent results and shifting task states. It utilizes a Dynamic Replanning Score (DRS) to measure how agents handle non-deterministic result arrivals and state updates through an Event Response Contract.
HOW THIS AFFECTS YOU
●
builderYou can use this benchmark to test how robust your agentic workflows are to asynchronous subagent delays.
●
researcherThis framework allows for more rigorous testing of agentic replanning capabilities beyond simple success metrics.