AI 用电脑，第一次有了全开源版本
GPT-6 Astra 的「用电脑」能力很强，但它是黑箱：你只知道结果，不知道它怎么观察、怎么决策。OpenGCUA 把整套系统开源了——从模型怎么选观察方式、怎么决定下一步，到怎么调用代码执行，全部公开。它最狠的一招叫「决策转移」:以前是固定规则决定 AI 看哪里、做什么，现在让模型自己决定。在 OSWorld-v2 基准上，这套配置让 8 个模型(包括 GPT-6 Astra 和 Claude Opus 5)的平均得分提升 19.6 分，完成任务所需步骤减少 53%。它还附带了真实 Mac 设备上的测试数据，发现 Qwen3.8-Max 的推理 token 消耗是 GPT-6 Astra 的 19.4 倍——同样是「用电脑」，效率差出两个数量级。这不是你明天能用的产品，但对研究者和想搞懂 AI 到底怎么操作电脑的人来说，这是第一次能拆开看内部构造的机会。
📄 原文摘要(英文)
Computer use has reached a turning point: the focus is expanding from GUI-only interaction to systems that complete tasks on users' computers through visual and programmatic interfaces. We define a generalist computer-use agent (GCUA) as a system in which the model chooses how to observe, act, and organize its work across visual and programmatic interfaces. Yet frontier implementations, such as GPT-6 Astra's computer-use system, remain largely closed, limiting systematic study of their design. We introduce Open Generalist CUA (OpenGCUA), an ongoing project to develop and open the full stack for GCUAs with the community. This report presents the first release of OpenGCUA. Code execution is a key feature of our agent design: it lets intermediate results determine subsequent operations and enables models to apply their existing programming capabilities directly to computer interaction. We also propose harness-to-model decision transfer, moving selected observations and task-state decisions from fixed harness rules to the model. On OSWorld-v2, our full configuration improves task scores across all eight evaluated models, including GPT-6 Astra and Claude Opus 5, raising mean partial score by +19.6 points and reducing reported steps per task by an average of 53.0% relative to batched GUI-only execution. We release our implementation and prompts, evaluation trajectories from all eight models, and preliminary versions of a real-device harness and a cross-device SDK. The harness addresses real-device challenges not fully captured by benchmarks, while the SDK provides a common programming interface across computers, mobiles and browsers. Across four real macOS tasks under our real-device harness, selected Qwen3.8-Max runs use 19.4 times as many reported reasoning tokens as GPT-6 Astra, highlighting substantial differences between models in real-device execution efficiency.