跳到主要内容Skip to main content

推理成本Inference Cost

Inference Cost

  • 选型与成本Selection & cost
  • 第四站 · 看懂Station 4 · Under the Hood
  • 引用核验 2026-11-03refs checked 2026-11-03
解决什么问题What problem it solves

模型每回答一次要花的钱,由输入和输出的 token 量决定。What a model costs per answer, determined by how many tokens go in and come out.

这是什么What it is

通常按输入 token 和输出 token 分别计价,输出更贵。

同一个模型,问法不同,账单可以差一个量级。

Usually priced separately for input and output tokens, with output more expensive.

The same model can cost an order of magnitude more depending on how you phrase the question.

为什么需要知道Why it matters

这是把 AI 用在日常场景时最容易踩的坑。它也是「选模型」必须考虑的成本项。The easiest trap to fall into when using AI day to day, and a cost item any model choice has to account for.

出处Sources
  • OpenAI 官方 API 定价OpenAI API Pricing ↗
  • OpenAI 官方模型文档OpenAI Models Documentation ↗