WHAT NEW IN X11.02 MON

X 上的 AI 合集档案 · 22 个合集 · 65,263 条

gpt-6POST

Clipping Bench:GPT-6 Astra 拿 300 美元跑到 9514 美元

给 AI 代理各 300 美元运营剪辑频道,GPT-6 Astra 以 9514 美元夺冠,但仍落后最优解 2.9 倍。

clay.
@clayul23 天前收录原推发布于 2026.10.10
clay. @clayul原推发布于 2026.10.10
在 X 上打开原推

The benchmarks I find most interesting aren't the ones with the hardest math. Real platforms. Real campaigns. Real money on the line. Clipping Bench gave AI agents $300 each to run clipping channels, and GPT-6 Astra finished first with $9,514. Even the winner was still 2.9x behind the oracle, so there's a lot of room left.

这条是引用着另一条写的:x.com/usenaive/status/2108592968596935013

x.com/clayul/status/2108603754773623236采集于 2026.10.10 10:03