Kimi K3.1 Pro Pricing Announced: $2 per Million Tokens Input, $10 Output

Damn! Absolutely explosive — Kimi just dropped a surprise release of its 2.8T flagship K3.1 Pro, taking first place on SWE Marathon with 51.7 and overtaking Opus 5.5 and GPT-6 Astra head-on! And this benchmark battle report that just came out has punched a huge hole through most of the closed-source large-model camp's defenses 🤯.. Officially, the company has rolled out its new multimodal flagship K3.1 Pro: total parameter count stacked all the way up to 2.8T, with a 1M native context window as standard. But in my view, the 2.8T figure may just be surface muscle; what's truly lethal in the whole announcement is two things: First, in the most valuable code and long-horizon agent tests, it goes head-to-head with Opus 5.5 and even overtakes GPT-6 Astra. Second, the company explicitly promises in black and white to fully open-source the complete weights before November 4. Many people just stare at the overall leaderboard for the spectacle when looking at benchmarks, but if you carefully break down this benchmark table, three key battles directly determine where engineering deployment goes next:

View original

Damn! 太炸裂了，Kimi 刚刚突袭发布的 2.8T 旗舰 K3.1 Pro，SWE Marathon 拿下 51.7 登顶第一，正面反超 Opus 5.5 与 GPT-6 Astra！ 而且刚刚放出的这张跑分战报，把整个大模型闭源阵营的防线打穿了一大半🤯。。 官方正式端出多模态新旗舰 K3.1 Pro：总参数量直接堆到了 2.8T，标配 1M 原生上下文窗口。 但在我看来，2.8T 这个数字或许只是表面肌肉，整篇公告里真正具备杀伤力的是两件事： 第一，在最具含金量的代码和长程智能体测试里，它正面咬住了 Opus 5.5，甚至反超了 GPT-6 Astra。 第二，官方直接白纸黑字承诺，11 月 4 日前把完整权重全盘开源。  很多人看跑分只盯着总榜看热闹，但仔细拆解这张基准表，有三场关键战役直接决定了接下来的工程落地走向：