Ted Sanders 解释 OpenAI 模型命名与规模逻辑
原标题：Quoting Ted Sanders
AI 摘要
OpenAI 的 Ted Sanders 解释了 LLM 性能的多维性，指出模型规模是解释性能差异的主要单一变量，更大的模型通常更可靠、每 token 成本更高、对冷门事实的召回更好。他同时说明了 OpenAI 的模型命名逻辑：Luna < Terra < Sol < Astra 的排序，以及版本号同时表示发布顺序、与其他版本的关系和改进幅度，例如 6.1 表示在 5.6 之后发布、与同系列 6.1 模型共享训练数据或风格，但提升幅度不足以称为 7。
正文节选
28th October 2026 In general, LLM performance is incredibly multidimensional. There's no guarantee of a fixed ordering where the model that is best at math is also the model that's best at jokes. Despite that complexity, much of the variation in performance is explained by just a single variable: model size. In general, models that are bigger will (a) be more performant / reliable, (b) cost more per token, and (c) have better recall of esoteric facts. As you note, our models are ordered: Luna <