Event ·Multi-source

GPT-5.6 Sol 在 ARC-AGI-3 基准上实现 SoTA,仅需两个设置变化

Sam Altman 转发消息称,OpenAI 的 GPT-5.6 Sol 在 ARC-AGI-3 基准测试中达到 SoTA(最高性能)。该模型仅通过两个设置变化实现:允许其在多个上下文窗口中进行推理,并配合规范化的压缩实现。这一结果暗示通过调整推理机制,模型在挑战性视觉推理任务上表现显著提升。目前官方博客已发布详细方法说明。

Current read

OpenAI 的 GPT-5.6 Sol 改了俩设置就在 ARC-AGI-3 上屠榜,教你多窗口推理+规范化压缩这套新打法。

11 sources63° AI heatLast updated 2026/7/30 00:41:57

Evidence

11 sources
Related source 4@OpenAIDevs
View source
Related source 6Amjad Masad
View source
Related source 7lmarena.ai
View source
Related source 8Greg Brockman
View source
Related source 9歸藏(guizang.ai)
View source
Related source 10SuperTechFans
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.