Claude Haiku 5.5 · DRACO · 100-task Perplexity deep-research benchmark · 成绩与配置

Claude Haiku 5.5 的 DRACO · 100-task Perplexity deep-research benchmark 记录。本次发布有 5 条成绩记录,含 5 条精确值(可含历史记录),附 1 条来源定位。

成绩与评测条件

Claude Haiku 5.5 · DRACO 成绩与配置
指标与成绩配置与评测条件出处
Normalized score64.3%原文 64.3% · 单位 %
公布 未确认
System Card effort / low
设置
Adaptive thinking; low effort, explicitly labeled in the source; benchmark-specific protocol is preserved in each result's conditions.
未知项
Sampling and trial counts vary by evaluation and are recorded per result.
工具
允许工具
推理预算
low
框架
未记录
协议
DRACO; 100 curated deep-research tasks; web search, web fetch, programmatic tool calling, code execution; 980k-token budget; five independent grader runs per response, averaged.
条件
low effort. Anthropic used an Opus 4.6 judge and a frozen offline web index; its score is not directly comparable to Perplexity’s headline score. Labeled figure value.
在排名中查看此成绩
Normalized score72.4%原文 72.4% · 单位 %
公布 未确认
System Card effort / medium
设置
Adaptive thinking; medium effort, explicitly labeled in the source; benchmark-specific protocol is preserved in each result's conditions.
未知项
Sampling and trial counts vary by evaluation and are recorded per result.
工具
允许工具
推理预算
medium
框架
未记录
协议
DRACO; 100 curated deep-research tasks; web search, web fetch, programmatic tool calling, code execution; 980k-token budget; five independent grader runs per response, averaged.
条件
medium effort. Anthropic used an Opus 4.6 judge and a frozen offline web index; its score is not directly comparable to Perplexity’s headline score. Labeled figure value.
在排名中查看此成绩
Normalized score77.8%原文 77.8% · 单位 %
公布 未确认
System Card effort / high
设置
Adaptive thinking; high effort, explicitly labeled in the source; benchmark-specific protocol is preserved in each result's conditions.
未知项
Sampling and trial counts vary by evaluation and are recorded per result.
工具
允许工具
推理预算
high
框架
未记录
协议
DRACO; 100 curated deep-research tasks; web search, web fetch, programmatic tool calling, code execution; 980k-token budget; five independent grader runs per response, averaged.
条件
high effort. Anthropic used an Opus 4.6 judge and a frozen offline web index; its score is not directly comparable to Perplexity’s headline score. Labeled figure value.
在排名中查看此成绩
Normalized score80.5%原文 80.5% · 单位 %
公布 未确认
System Card effort / xhigh
设置
Adaptive thinking; xhigh effort, explicitly labeled in the source; benchmark-specific protocol is preserved in each result's conditions.
未知项
Sampling and trial counts vary by evaluation and are recorded per result.
工具
允许工具
推理预算
xhigh
框架
未记录
协议
DRACO; 100 curated deep-research tasks; web search, web fetch, programmatic tool calling, code execution; 980k-token budget; five independent grader runs per response, averaged.
条件
xhigh effort. Anthropic used an Opus 4.6 judge and a frozen offline web index; its score is not directly comparable to Perplexity’s headline score. Labeled figure value.
在排名中查看此成绩
Normalized score81.5%原文 81.5% · 单位 %
公布 未确认
System Card effort / max
设置
Adaptive thinking; max effort, explicitly labeled in the source; benchmark-specific protocol is preserved in each result's conditions.
未知项
Sampling and trial counts vary by evaluation and are recorded per result.
工具
允许工具
推理预算
max
框架
未记录
协议
DRACO; 100 curated deep-research tasks; web search, web fetch, programmatic tool calling, code execution; 980k-token budget; five independent grader runs per response, averaged.
条件
max effort. Anthropic used an Opus 4.6 judge and a frozen offline web index; its score is not directly comparable to Perplexity’s headline score. Labeled figure value.
在排名中查看此成绩

出处与披露

测试版本与指标

别名 Deep Research Accuracy, Completeness, and Objectivity

  • Normalized score · % · 越高越好