Finance & Accounting Index
评估模型在财务与会计领域的表现。评估的能力包括会计、投资、企业与市场等领域知识，以及财务分析与报告、合规与审计、市场研究等。
查看代表性工作流The Artificial Analysis Finance & Accounting Index combines performance across Intelligence benchmarks sliced for finance and accounting tasks. We map common tasks from O*NET occupational classifications, then select benchmarks that represent this real-world work. Weights are derived from how often capabilities appear across those tasks.
This composite metric provides a single score for tracking model performance across finance and accounting tasks. 所有底层基准测试均由 Artificial Analysis 独立运行。有关评测的实施方式，请参阅我们的智能基准测试方法论。
| 能力 | 权重 | 评测 |
|---|---|---|
| 商业知识 | 30% | AA-Omniscience 商业准确率 |
| 智能体知识工作 | 30% | GDPval-AA v2.1和AA-Briefcase v1.1 |
| 推理 | 20% | HLE |
| 智能体工具使用 | 10% | AutomationBench-AA 财务 |
| 长上下文 | 5% | LCR和GDP.pdf |
| 避免幻觉 | 5% | AA-Omniscience 商业避免幻觉 |
AA-Omniscience: Knowledge and Hallucination BenchmarkGDPval-AA v2.1 LeaderboardAA-Briefcase v1.1: Agentic Knowledge Work Benchmark
Humanity's Last Exam Benchmark Leaderboard
AutomationBench-AA: Agentic SaaS Workflow BenchmarkArtificial Analysis Long Context Reasoning Benchmark Leaderboard
GDP.pdf Benchmark Leaderboard得分
Artificial Analysis Finance & Accounting Index
Artificial Analysis Finance & Accounting Index：能力明细
能力明细
Artificial Analysis Finance & Accounting Index：商业知识
代表性工作流
这些真实工作流重点检验 Finance & Accounting Index 中权重最高的能力。
示例：Build a same-day EBITDA bridge for an acquisition target from five years of audited 10-Ks and an unaudited Q3 pack to reconcile GAAP and IFRS treatment, surface customer-concentration risk in the margin walk, and assemble a trading-comps table.
示例：Reconstruct twelve months of undocumented expense reimbursements from the general ledger and journal entries ahead of a SOX audit to tie each line to source evidence, produce an auditable trail, and flag unsupported items.
示例：Recover a twice-re-scoped SAP S/4HANA rollout where three departments are cross-blocked to map the critical-path dependencies, re-baseline milestones with explicit trade-offs, and draft stakeholder updates that state what slips if scope stays fixed.
示例：Reconcile two vendor studies reporting opposite consumer preferences for the same launch to compare their sampling and conjoint methodologies, explain plausible reasons for the split, and recommend the lower-risk go-to-market path.
示例：Diagnose a fulfilment centre whose pick error rate doubled after a floor reorganisation to rank likely root causes from shift logs and layout data, and propose interventions by expected lift.
成本
Artificial Analysis Finance & Accounting Index：单任务成本
Artificial Analysis Finance & Accounting Index 与单任务成本
速度
Artificial Analysis Finance & Accounting Index：单任务耗时
输出 token
Artificial Analysis Finance & Accounting Index：单任务输出 token
发布日期
Artificial Analysis Finance & Accounting Index 与发布日期
常见问题
根据 Artificial Analysis Finance & Accounting Index，目前在金融与会计工作上表现最佳的 AI 模型是 Claude Opus 5.5 (Max, Default Fallback) (61)、Claude Opus 5.5 (Xhigh, Default Fallback) (58)和Claude Sonnet 5.5 (Max, Default Fallback) (57)。新模型发布后，排行榜会随之更新。
有。Artificial Analysis Finance & Accounting Index 是一项独立基准测试，用于衡量 AI 模型在金融与会计工作上的表现。它评估商业知识、智能体知识工作、推理、智能体工具使用、长上下文分析和避免幻觉等能力。
Finance & Accounting Index 是 Artificial Analysis 推出的综合基准测试，用于评估模型在金融与会计领域的表现。评估能力包括会计、投资、企业与市场等专业知识，以及金融分析与报告、合规与审计、市场研究等。
Finance & Accounting Index 按各项能力子分数的加权平均值计算。各项子分数及其权重为：商业知识 (30%)、智能体知识工作 (30%)、推理 (20%)、智能体工具使用 (10%)、长上下文 (5%)和避免幻觉 (5%)。
Finance & Accounting Index 包含 AA-Omniscience 商业准确率、GDPval-AA v2.1、AA-Briefcase v1.1、HLE、AutomationBench-AA 财务、LCR、GDP.pdf和AA-Omniscience 商业避免幻觉。
在已公布结果的模型中，Claude Opus 5.5 (Max, Default Fallback) 目前以 61 分位居 Finance & Accounting Index 榜首。 查看模型
Finance & Accounting Index 得分越高，表示模型在构成该指数的各项基准测试中整体表现越强。对于特定用例，单项基准测试结果可能比综合得分更具参考价值。