跳到正文
𝕏 Rohan Paul(x.com)·· 5 天前品質評分53
重點導讀

估算大型語言模型耗能時,如果只計算每次產生的新文字,可能會漏掉 AI 助理反覆讀取舊內容的成本。

  • 研究作者記錄 1 位工程師使用 Claude Code 10 週的情況,估算每產生 1 個 token,約讀取 260 個快取 token。把這些讀取也算進去後,估計耗能約 250 kWh。
  • 只按輸出文字估算則約 10 kWh。作者建議追蹤 AI 助理的成本或排放時,也計入快取讀取與子助理呼叫,並盡可能用新的工作階段開始新任務。

AI 伴隨問答

自備 Key 深度分析

針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:

Rohan Paul
Rohan Paul@rohanpaul_ai
·5 天前
𝕏

New Harvard paper finds that 94.8% of a coding agent's tokens were re-reads of cached context, i.e. 94.8% of the text the AI processed was earlier conversation it was re-reading, not new work.

And that re-reading raises its estimated energy use about 25 times, from roughly 10 to 250 kilowatt-hours over 10 weeks.

Energy estimates for LLMs usually price single request-and-response calls. Agents re-read a growing context on every call and hand work to subagents, which those estimates can miss.

The authors logged 10 weeks of real Claude Code use by 1 engineer, about 1 billion tokens across 42 sessions. They built an energy estimator that charges cached tokens for attention reads but not full reprocessing.

About 260 cached tokens were read per generated token, and context grew 5.3× over a session. The cache-aware estimate came to about 250 kWh, against about 10 kWh from the output-based EcoLogits estimator.

If you track agent cost or emissions, count cached reads and subagent calls, and start new tasks in fresh sessions when possible.

推文來源 · X (Twitter)在 X 參與討論

讀者討論

理性交流 · 即時討論