gpt-6POST
Kimi K3.1 Pro 在 10/14 项基准上胜过 Astra
爆料称 Kimi K3.1 Pro 在 14 项基准中赢 Opus 5.5 九项、赢 GPT-6 Astra 十项,SWE Marathon 得分 51.7 对 Astra 的 42.5。
Haider.
@haider15 天前收录原推发布于 2026.10.24
Haider. @haider1原推发布于 2026.10.24
okay, wow. kimi k3.1 pro beats opus 5.5 on 9/14 benchmarks and gpt-6 astra on 10/14 and even on SWE Marathon kimi scores 51.7 vs 46.3 for opus and 42.5 for astra didn't expect kimi to be competing this closely with frontier models
这条是引用着另一条写的:x.com/kimi_moonshot/status/2113680701770068067
x.com/haider1/status/2113682109164622316采集于 2026.10.24 10:09