gpt-6IMAGE
GPT-6 Astra 做满 FrontierMath Tier 4
Epoch 的 Tier 4 基准从 5% 起步,Astra 解掉了最后一道题,该档位已饱和。
Andrew Curran
@AndrewCurran_2026.09.14收录原推发布于 2026.09.11

Andrew Curran @AndrewCurran_原推发布于 2026.09.11
When Epoch launched FM Tier 4 on July 11th, 2025, the top score was 5%. GPT-6 Astra has now solved the last problem standing. FrontierMath Tier 4 is saturated.
这条是引用着另一条写的:x.com/epochairesearch/status/2098103831502708864
x.com/andrewcurran_/status/2098132654109732917采集于 2026.09.14 15:41