PulseAugur
中
实时 23:12:07
English(EN) Why Your Coding Agent Costs 5x More Than It Should (It Is Not the Model)

编码代理成本由模型驱动,但工具链会增加费用

一项新的基准测试 FrontierHarness Eval v1.1 显示,虽然 AI 模型本身占编码代理成本变化的 92.8%,但工具链(harness)配置可以将相同模型的成本提高高达 5 倍。研究发现,很大一部分支出(64.5%)用于最终失败或超时的试验,代理经常提供自信但错误的最终答案。这表明 AI 编码代理的成本更多地受到所采取的步骤数量和验证过程的影响,而不仅仅是模型的能力。 AI

影响 强调 AI 代理的成本很大程度上受工具链配置和验证的影响,而不仅仅是模型本身,这会影响运营效率。

排序理由 发布新的基准测试,对 AI 代理性能和成本因素进行详细分析。[lever_c_demoted from research: ic=1 ai=1.0]

在 dev.to — LLM tag 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

编码代理成本由模型驱动,但工具链会增加费用

本文如何被排名

Signal score
57 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
发布新的基准测试,对 AI 代理性能和成本因素进行详细分析。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
infra, product
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Felipe Castillo ·

    为什么你的编码代理花费比应有的多 5 倍(不是模型的问题)

    <p>Runta published FrontierHarness Eval v1.1 today: 27,280 agent runs, 83 billion tokens, 22 harness configurations, 5 models, 62 terminal tasks. It is the largest public comparison of coding agents I have seen, and it is free to read.</p> <p>I read it expecting one answer and fo…