AstronOS: A Unified Execution Model and Runtime for Long-Horizon Agentic Systems
2026-08-17 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors propose a way to keep track of work that happens in many steps, by maintaining a consistent and versioned state throughout. They test five methods to carry out software update plans across multiple steps, including a new method using their system called AstronOS. Their tests show AstronOS performs much better at completing these multi-step tasks correctly compared to the other methods. However, AstronOS takes a bit more time per attempt, though it uses fewer tokens in the model. Overall, the authors find their approach leads to more successful, continuous sessions despite some time cost.
agentic systemsexecution modelpersistent stateversioningsoftware update planAstronOSmulti-step tasksmodel token costdeterministic summarybenchmarking
Authors
Zhenhang Nie, Gui Zheng, Xudong Sun, Tailong Zhu, Bin Zhang
Abstract
Agentic systems often organize execution and state around a single conversation, model invocation, or agent instance, even when real work spans many calls and stages. We introduce a unified execution model that maintains a work item's persistent identity and versioned authoritative state across calls. Each step receives input scoped to a specific state version and new material; a result advances state only after validation and recording. We implement selected paths of this model in AstronOS using Cases, Tasks, and Scenario Packs across central and local execution. We compare five complete strategies for carrying an established software-version update plan into a fresh model session: rereading original materials, replaying full history, deterministic text summary, deterministic JSON, and the AstronOS runtime-mediated handoff. Ten controlled tasks are run under all five strategies with three repetitions, yielding 150 included executions. On the single-stage reference family, strategies perform similarly. In the primary three-stage A-C batch, AstronOS passes the frozen scorer in 14 of 15 executions, compared with 0 of 15 for rereading and 2 of 15 for full-history replay; later non-interleaved summary and JSON batches each pass 0 of 15. AstronOS has lower attempt-accounted model-token cost per passing execution, while requiring more execution-window time per attempt. These results associate the complete AstronOS condition with higher end-to-end pass rates across fresh sessions in this benchmark, at a measurable time cost.