---
id: 20261002-T0-07
title: "AREX-2：通过长周期反思任务提升LLM Agent自我改进能力"
title_en: "AREX-2 Advances Self-Improving LLM Agents via Long-Horizon Reflective Tasks"
url: https://ai.daily.yangsir.net/daily/20261002-T0-07
issue_date: 2026-10-02
publish_date: 2026-10-01T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2609.38288
---

# AREX-2：通过长周期反思任务提升LLM Agent自我改进能力

研究人员提出AREX-2，旨在提升LLM Agent的自我改进能力，即在测试时迭代优化解决方案的能力。该方法依赖两项核心能力：反思（生成对当前方案的评估和改进方向）和长周期任务执行。通过让Agent在长周期任务中反复反思和调整，AREX-2在自我改进 benchmark 上取得了进展。这为构建能在部署后持续进化的AI Agent提供了新思路。

## English Version

**AREX-2 Advances Self-Improving LLM Agents via Long-Horizon Reflective Tasks**

Researchers present AREX-2, an effort to advance self-improving LLM agents—the ability to iteratively refine solutions at test time. The method rests on two capabilities: reflection, which produces solution assessments, and long-horizon task execution. By having agents repeatedly reflect and adjust during extended tasks, AREX-2 makes progress on self-improvement benchmarks, offering a new path toward AI agents that continue to evolve after deployment.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2609.38288)

**详情页**：https://ai.daily.yangsir.net/daily/20261002-T0-07

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*