Event ·Multi-source

Anthropic发布Haiku 5.5在ARC-AGI评测中表现优异

Anthropic发布了小模型Haiku 5.5,在ARC-AGI-1基准测试中达到89.0%的准确率,任务成本为0.051美元。在ARC-AGI-2测试中,Haiku 5.5准确率达到61.3%,每任务成本0.085美元。在最新的ARC-AGI-3测试中,使用provider adapter harness获得2.8%的准确率,成本437美元;使用standard harness获得0.5%的准确率,成本388美元。

Current read

Anthropic小模型Haiku 5.5在ARC-AGI评测中表现亮眼,性价比高

11 sources61° AI heatLast updated 2026/10/14 17:10:03

Evidence

11 sources
Related source 1Teortaxes
View source
Related source 2量子位
View source
Related source 3Artificial Analysis
View source
Related source 5Tech in Asia
View source
Related source 6TestingCatalog
View source
Related source 7kimmonismus
View source
Related source 8Ethan Mollick
View source
Related source 10OpenRouter
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.