产品40
OpenAI 新工具检测推理前缀编辑
原标题:Adjusting an agent’s reasoning effort or adding tools can sometimes lead to unintended prefix edits, depending on how those changes are a...
调整智能体的推理强度或添加工具,有时会导致意外的前缀编辑,具体取决于这些更改的应用方式。 为帮助开发者检测这一问题,我们构建了新工具,例如 Responses API 中的“input_transformations”对象,以及 Usage 仪表板中新增的“Reasoning removals”部分。
中文译文尚未完成,先为你展示原文。
Adjusting an agent’s reasoning effort or adding tools can sometimes lead to unintended prefix edits, depending on how those changes are applied.
To help developers detect this, we’ve built new tools like the “input_transformations” object in the Responses API, and the new ‘Reasoning removals’ section in the Usage dashboard.
