Models

Epoch AI:GPT-6.2 Astra 延续 9 月以来的新推理基准趋势线

Why we picked it

Epoch AI 给 GPT-6.2 Astra 的基准成绩画了条趋势线,结论是 9 月之后推理模型换了条更陡的增长曲线。做 AI 产品排期的开发者可以看看这套追踪方法,估能力上限时别再按旧斜率外推了。

Epoch AI 的分析指出,OpenAI 的 GPT-6.2 Astra 在推理基准上的表现延续了自 9 月以来的上升趋势线,并且已经越过所谓旧推理时代(old-world reasoning era)的水平线。发布该观点的 Andrew Curran 认为,这支持了他此前的判断:从 9 月开始,模型能力进入了一条新的增长轨迹。

Body

从机制上看,这类判断依据的是 Epoch AI 对前沿模型基准成绩的纵向追踪:把各代模型在同一基准上的得分连成趋势线,观察是否出现斜率变化。GPT-6.2 Astra 的成绩落在新趋势线之上,说明其提升幅度超出旧时代的基线外推。

对从业者的意义在于,如果这条新趋势线成立,前沿推理模型的代际提升节奏可能与 2023-2024 年的旧节奏不同,开发者在规划产品能力和成本预期时需要按新的提升斜率来估算。

Original · Andrew Curran

Epoch says GPT-6.2 Astra continues the trendline above the old-world reasoning era, supporting my argument that we have been on a new trajectory since September, which will eventually become visible and undeniable. https://t.co/eFagaXbreH

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include GPT-6.2, Epoch AI, OpenAI, reasoning models.

Source attribution: curated from Andrew Curran (@AndrewCurran_).