ProductsSelected

Perplexity 公布工具调用改进数据:失败率下降 11%

Why we picked it

Perplexity 晒了自家 agent 的实测数据:工具调用失败率降 11%,用户不满线程降 7.6%,成本和延迟没变。做 agent 的团队可以拿这组口径对照自己产品的稳定性指标,比看跑分实在。

Perplexity 发布了一组内部测试数据:在相同测试条件下,工具调用失败率在全部 Computer activity 上下降了 11%,剔除系统自身后台运行后约下降 7%。用户体验层面的指标同步改善,每位用户的中、高不满度对话线程减少了 7.6%。

Body

从细节看,这次改进集中在工具调用链路的稳定性上。Perplexity 特意区分了包含后台运行和剔除后台运行两种口径,说明部分失败来自系统自动执行的任务,而非用户直接触发的调用。值得注意的是,成本和响应时间在改进前后没有可测量的变化,意味着稳定性提升没有以额外的延迟或费用为代价。

这类公开失败率与不满度指标的做法在 agent 类产品中并不多见,多数厂商只公布基准分数。对开发者和企业用户而言,工具调用可靠性是决定 agent 能否用于生产环境的关键变量,这组数据提供了可横向参考的基线。

Original · Perplexity

In the same test, the tool call failure rate dropped by 11% across all Computer activity, and by about 7% excluding the system's own background runs. Medium- and high-dissatisfaction threads per user dropped by 7.6%. Cost and response time didn't measurably change.📊 2 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Perplexity, tool use, agent, agents.

Source attribution: curated from Perplexity (@perplexity_ai).