Kimi K3.1 Pro 在 Terminal-Bench 4.0 上超过 GPT-6 Astra

Kai @hqmank原推发布于 2026.10.24

在 X 上打开原推

Kimi K3.1 Pro is getting remarkably close to the top closed models in coding. On Terminal-Bench 4.0, it's just 0.7 points behind Claude Opus 5.5. It also outperforms GPT-6 Astra and Fable 5.1. That's impressive for a model going open-weight. The gap is getting smaller. Now I want to see if it can get this close to Opus on real coding projects.

这条是引用着另一条写的：x.com/kimi_moonshot/status/2113680701770068067