OpenAI 发布 GPT-6 Astra，首席研究官 Mark Chen 称其能构建测试软件、跨应用操作电脑并尝试开放科学问题。
作者补充数字：OSWorld 真实桌面任务从 75 分钟降到 40 分钟，职场自动化从 18% 升到 41%，未防护越权率从上一代 48% 压到 0%，并称花了 2000 美元算力解出 10
道十年未解的数学与理论计算难题；
同时指出带工具的综合测试仍落后 Claude。
AI News / Industry
Back to AI news
OpenAI released GPT-6 Astra, focusing on computer operation and alignment capabilities
OpenAI released GPT-6 Astra, with Chief Research Officer Mark Chen claiming it can build testing software, operate computers across applications, and exper...
Today AI Intelligence Brief
OpenAI released GPT-6 Astra, with Chief Research Officer Mark Chen claiming it can build testing
software, operate computers across applications, and experiment with open science problems. The author adds figures: OSWorld's real desktop tasks dropped from 75 minutes to 40 minutes, workplace automation increased from 18% to 41%, unprotected overreach rate dropped from 48% in the previous generation to 0%, and it was said that $2,000 of computing power was spent to solve 10 mathematical and theoretical problems that had remained unsolved in ten years; At the same time, it pointed out that comprehensive tests with tools still lag behind Claude.
动态正文 · AI 整理
Intelligence Assessment
Aioga 编辑摘要
来源材料称，OpenAI 发布 GPT-6 Astra，首席研究官 Mark Chen 表示该模型能够构建测试软件、跨应用操作电脑，并尝试处理开放科学问题。
背景分析
材料还提到，作者补充了 OSWorld 桌面任务用时、职场自动化比例及未防护越权率等对比数据，并称其以 2000 美元算力完成 10 道数学与理论计算难题；同时指出带工具的综合测试仍落后 Claude。
Aioga 观点
Aioga 判断：目前材料同时包含发布信息、作者补充数据和对比结论，但未提供完整测试方法、评测条件或独立验证，因此不宜仅凭这些内容确认模型能力的普遍性。
影响与后续
可能影响：若相关能力和数据经后续核验，GPT-6 Astra 可能加强电脑操作与科研辅助方向的竞争；但单一来源信息不足以证明其已稳定适用于职场或开放科学场景。 后续观察：需要关注 OpenAI 是否公开正式技术报告、评测口径、工具使用条件和安全测试细节，并核对作者所述数据及其与 Claude 对比的具体范围。
Source and Copyright
The readable text on this page was extracted from the public source and organized with attribution, publication time and the original link. Copyright remains with the original author and publisher.
Ingestion channel: Summary aggregation · Source domain: x.com
Source: @AYi_AInotes)
Original link: Open original source
Aioga archive: Open intelligence page
Content record: social-summary · Updated: 2026-09-04T02:22:24.000Z
API RELAY · DEVELOPER INFRASTRUCTURE
API 中转站
统一接入主流 AI 模型 API，为开发、测试与生产环境提供稳定调用入口。
立即访问 api.w173.com
Scan to open this article