Effort and run count are not stated in the comparison table; linked methodological details were not independently reviewed in this task. 来源主文最初发布日期为2026-02-05,但当前HTML修改于2026-10-14;没有与此数值匹配的首发图像历史字节,因此精确首次公布日未知。
Artificial Analysis independent evaluation; 220 GDPval tasks across 44 occupations and 9 industries; agent has shell and web browsing; blind pairwise output comparisons.
未知项
未另记
工具
允许工具
推理预算
未记录
框架
未记录
协议
GDPval-AA independent Elo evaluation.
条件
Artificial Analysis evaluation over 220 GDPval tasks, 44 occupations and 9 industries; shell and web access; blind pairwise comparisons.
Opus 4.8 defaults to high effort; table values follow the model's reported score. OSWorld-Verified uses an updated evaluation described in the release footnote.
未知项
Per-row sampling and trial counts are not stated in the main table.
工具
允许工具
推理预算
high
框架
未记录
协议
GDPval-AA; 220 tasks across 44 occupations and 9 industries; shell and web browsing in an agent loop; Elo from blind pairwise comparisons.
条件
Evaluation run independently by Artificial Analysis; preserve the release-table value of 1890 Elo.
Differences are calculated only for unique, active, exact values between direct predecessors when the unit, metric, and explicitly recorded configuration and conditions match.
Effort and run count are not stated in the comparison table; linked methodological details were not independently reviewed in this task. 来源主文最初发布日期为2026-02-05,但当前HTML修改于2026-10-14;没有与此数值匹配的首发图像历史字节,因此精确首次公布日未知。
· 1 score record for this release, including 1 exact value (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Claude Opus 4.7 · GDPval-AA scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Elo rating1753 EloReported 1,753 Elo · Unit Elo
Published Unconfirmed
GDPval-AA独立Elo评测
Settings
Artificial Analysis independent evaluation; 220 GDPval tasks across 44 occupations and 9 industries; agent has shell and web browsing; blind pairwise output comparisons.
Unknowns
none recorded
Tools
Tools allowed
Reasoning budget
not recorded
Framework
not recorded
Protocol
GDPval-AA independent Elo evaluation.
Conditions
Artificial Analysis evaluation over 220 GDPval tasks, 44 occupations and 9 industries; shell and web access; blind pairwise comparisons.
· 1 score record for this release, including 1 exact value (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Claude Opus 4.8 · GDPval-AA scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Elo rating1890 EloReported 1890 Elo · Unit Elo
Published Unconfirmed
发布表报告值
Settings
Opus 4.8 defaults to high effort; table values follow the model's reported score. OSWorld-Verified uses an updated evaluation described in the release footnote.
Unknowns
Per-row sampling and trial counts are not stated in the main table.
Tools
Tools allowed
Reasoning budget
high
Framework
not recorded
Protocol
GDPval-AA; 220 tasks across 44 occupations and 9 industries; shell and web browsing in an agent loop; Elo from blind pairwise comparisons.
Conditions
Evaluation run independently by Artificial Analysis; preserve the release-table value of 1890 Elo.