模型83°
ARC-AGI基准测试成本大幅下降
精选理由
DeepSeek V4 Flash和Dots3-Note Preview在ARC-AGI基准测试中实现了成本大幅下降，比之前的模型便宜数千倍。
在ARC-AGI-1基准测试中，19个月内达到75%准确率的成本下降了99.95%，从o3-preview的26美元/任务降至DeepSeek V4 Flash的0.01美元/任务。在ARC-AGI-2基准测试中，仅6个月内成本下降了99.44%，从Gemini 3 Deep Think的13.62美元/任务降至Dots3-Note Preview的0.08美元/任务。
原文 · ARC Prize
Over the course of 19 months, the cost to reach 75% on ARC-AGI-1 fell 99.95%: from o3-preview at $26/task to DeepSeek V4 Flash at $0.01/task.
On ARC-AGI-2, it fell 99.44% in just 6 months: from Gemini 3 Deep Think at $13.62/task to Dots3-Note Preview at $0.08/task. https://t.co/2tmu15pjQ3
- arXiv: DeepSeek09-30 05:23原文