Simular Agent S4：用元智能体自动演化智能体操作界面
Simular 这篇论文挺有意思：让一个元智能体自动给电脑操作智能体造界面和工具，27B 开源模型干赢了 GPT-5.5，成本还便宜几十倍。做智能体的都该看看。
Simular 的 Agent S4 引入一个元智能体，读取任务智能体的执行轨迹（无需成功标签），自动编写新工具及使用说明，替代人工设计的操作界面。在 OSWorld V2 基准上，一个开源 27B 模型配合演化后的界面达到 21.3%，是手动界面的 3.8 倍，超过 GPT-5.5（13.0%）和 Claude Opus 4.8（20.6%），成本仅约为两者的 4.3% 和 1.5%。通过基于种群的筛选机制，界面仓库在元智能体未见过的验证任务上依然有效。在 ScienceBoard 基准上，同一模型的成功率从 41.4% 提升到 58.6%，代码已开源。
Banger paper on evolving agent interfaces.
Pay close attention to this new trend if you build with agents.
Like optimizing harnesses, it's remarkable what you can achieve with agents that have access to better interfaces.
Computer-use agents usually act through interfaces that people design by hand, so an agent may type a spreadsheet cell by cell when one line of code would do. Simular's Agent S4 has a meta-agent that builds those interfaces instead. It reads the task agent's trajectories, with no success labels, and writes new tools and notes on when to use them.
With the evolved interfaces, an open 27B model reaches 21.3% on OSWorld V2, 3.8 times that of the same model with hand-built interfaces.
That is above GPT-5.5 (13.0%) and Claude Opus 4.8 (20.6%), at about 4.3% and 1.5% of their cost.
Population-based selection keeps the interface repositories that do best on validation tasks the meta-agent never sees, so the changes carry over to new tasks.
On ScienceBoard, the evolved interfaces raise success from 41.4% to 58.6% with the same model. The code is open source.
https://t.co/GugEMxwuzg
Chat with Paper: https://t.co/5stmPhni6v
