Event ·Official first-hand

APEX：用智能体辅助生产排程的可扩展框架

论文提出 APEX，一个围绕通用模型与混合多目标搜索构建的生产排程框架。智能体负责数据准备、自然语言场景探索，编码智能体则通过扩展钩子实现新的约束与目标。基准测试在 69 个公开 job-shop、flexible job-shop 和 permutation flow-shop 实例上，与 NSGA-II、SPEA2、MOEA/D、SMS-EMOA 四种算法对比，混合配置取得最佳总体解质量。另一项工作流研究用 OpenAI 的 GPT-6-astra 作为人机交互层，24 个会话全部完成规则添加、目标和计划修改并通过独立检查，编码评估产出 6 个新目标或硬约束的原生实现，全部通过验证且未改动核心。

Current read

生产排程这事儿可以用智能体来做了，APEX 让你用自然语言改约束、改目标，还跑赢了 NSGA-II 这些老牌算法，搞运筹优化的可以看看。

11 sources34° AI heatLast updated 2026/9/27 10:24:59

Evidence

11 sources

Primary sourcearXiv: OpenAI

2026/9/27 10:24:59View source

Related source 1@OpenAIDevs

2026/9/25 16:54:29View source

Related source 2AI Will

2026/9/26 02:39:58View source

Related source 3Bindu Reddy

2026/9/28 15:00:08View source

Related source 4kimmonismus

2026/9/28 19:47:20View source

Related source 5AIGCLINK

2026/9/29 03:06:48View source

Related source 6Aravind Srinivas

2026/9/25 14:24:21View source

Related source 7theverge

2026/9/25 15:39:48View source

Related source 8Thomas Wolf

2026/9/25 15:42:34View source

Related source 9techcrunch

2026/9/25 15:48:14View source

Related source 10The Information

2026/9/25 16:30:08View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story