Moonshot AI launches Kimi K3.1 Pro with 1M-token context
Moonshot AI released Kimi K3.1 Pro, a post-trained version of Kimi K3 with 2.8 trillion total parameters, 104 billion active parameters, and a 1 million-token context window. Artificial Analysis reported a score of 55 on its Intelligence Index, while the model is available through Kimi web and mobile services, Kimi Work, Kimi Code, the API, and OpenRouter. Moonshot AI said it would release the full model weights by November 4, but the supplied reports do not confirm that release occurred.
Kimi K3.1 Pro is a significant open-weight flagship release focused on coding, agents, 3D creation, and professional workflows, although the promised full-weight release remains unconfirmed.
过去两个月，全球大模型市场迎来一轮密集更新。
9 月，OpenAI 接连推出 GPT-6 Astra、GPT-6 Sol 等新模型，Anthropic 也发布了 Claude Opus 5.5 和 Claude Sonnet 5.5。
相比之下，中国头部大模型厂商的新一代旗舰模型发布节奏相对平静。自今年 7 月 Kimi K3 发布以来，已经有三个多月没有推出同等规模的旗舰升级。
10 月 23 日，月之暗面正式发布 Kimi K3.1 Pro，重新加入这场竞争。新模型拥有 2.8 万亿总参数，每次推理激活 1040 亿参数，支持 100 万 token 上下文窗口，在编程、智能体、3D 创作和金融研究等领域实现升级。官方宣布，完整模型权重将在 11 月 4 日前开放。
这次发布最重要的依旧是性能成绩。在月之暗面公布的 Terminal-Bench 4.0 编程测试中，K3.1 Pro 获得 65.7% 的成绩，超过 GPT-6 Astra 的 58.2%，距离 Claude Opus 5.5 的 66.4% 仅差 0.7 个百分点。
#Moonshot AI#Kimi K3.1 Pro#Kimi K3.1 Pro Highspeed#Artificial Analysis#OpenRouter#Kimi AI#Kimi Work#Kimi Code#Kimi API#3D Make#GPT-6 Astra#Claude Opus 5.5