跳到正文
𝕏 Artificial Analysis(x.com)·· 19 天前品質評分39
重點導讀

Artificial Analysis 將六大 Capability Indexes 升級至 v2.0,對齊 Artificial Analysis Intelligence Index v5.0 評測,並提高對 Non-Hallucination 與 Agentic Tool Use 的重視。

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:

Artificial Analysis
Artificial Analysis@ArtificialAnlys
·19 天前
𝕏

We’ve updated the six Artificial Analysis Capability Indexes to v2.0, aligning them with evaluations from the Artificial Analysis Intelligence Index v5.0 and placing greater emphasis on Non-Hallucination and Agentic Tool Use

The Capability Indexes map tasks from O*NET occupations to benchmarks that represent them, weighting each benchmark by how often its capability appears in that occupation. There are six indexes: Finance & Accounting, Strategy & Ops, Legal, Healthcare & Medical, Engineering, and Economics.

Key changes:

➤ Finance & Accounting: increased weight on Non-Hallucination, upgraded HLE to HLE-Diamond and AA-Omniscience to v1.1, removed GDPval-AA from Agentic Knowledge Work, and removed AA-LCR from Long-Context

➤ Strategy & Ops: increased weight on Agentic Tool Use, added Non-Hallucination and upgraded AA-Omniscience to v1.1, removed GDPval-AA from Agentic Knowledge Work, and removed AA-LCR from Long-Context

➤ Legal: increased weight on Non-Hallucination, upgraded HLE to HLE-Diamond and AA-Omniscience to v1.1, removed GDPval-AA from Agentic Knowledge Work, and removed AA-LCR from Long-Context

➤ Healthcare & Medical: increased weight on Agentic Tool Use, upgraded HLE to HLE-Diamond and AA-Omniscience to v1.1, and removed GDPval-AA from Agentic Knowledge Work

➤ Engineering: added Science as a new capability sourced from CritPt v2 (Hard Subset) and Terminal-Bench-Science 0.1, added FrontierCode 2.0 to Agentic Coding, upgraded HLE to HLE-Diamond and AA-Omniscience to v1.1, and removed GDPval-AA from Agentic Knowledge Work

➤ Economics: increased weights on Agentic Knowledge Work and Long-Context Reasoning, replaced AA-LCR with GDP.pdf in Long-Context Reasoning, upgraded HLE to HLE-Diamond and AA-Omniscience to v1.1, and removed GDPval-AA from Agentic Knowledge Work

Key results:

➤ Claude Opus 5.5 (max) leads in all six indexes, followed by GPT-6 Astra (max) in Finance & Accounting, Legal, and Engineering

➤ Among open weights models, Kimi K3.1 Pro (weights not yet released) leads all six indexes, at #5 to #8. MiniMax-M3.1-Flash leads in Strategy & Ops (#10), and DeepSeek V4.1 Flash 1017 (max) leads Healthcare & Medical (#12)

推文來源 · X (Twitter)在 X 參與討論

讀者討論

理性交流 · 即時討論