Moonshot AI releases Kimi K3.1 Pro with 3-trillion-level parameters and a 1-million-token context
Moonshot AI has released Kimi K3.1 Pro, a new-generation open-source large model with 3-trillion-level parameters that supports a 1-million-token context window, retaining its title as the open-source model with the largest parameter scale in the world. The model is a post-trained version of Kimi K3, with a substantial increase in post-training compute, and it shows notable improvements in coding, 3D creation, financial research and long-horizon Agent capabilities; across multiple international benchmarks, its overall intelligence level is now approaching that of the world's leading closed-source models. The report also noted that OpenRouter statistics show global total AI large-model API calls from October 12 to 18 reached 164 trillion tokens, and that the weekly call volume of Chinese large models has exceeded that of the United States for 25 consecutive weeks.
Body · AI translation
In the early hours of October 24, Chinese artificial intelligence (AI) large-model company Moonshot AI released its new-generation open-source large model Kimi K3.1 Pro. The model has a parameter scale of 3 trillion and supports a 1 million token context window, retaining its title as the open-source model with the largest parameter scale in the world.
Kimi K3.1 Pro is a post-training version of Kimi K3, with a substantial increase in post-training compute. In July this year, Moonshot AI released the new-generation model Kimi K3, with a parameter scale of 2.8 trillion. A larger parameter scale means the model can pack more knowledge and patterns into its "brain," know more, think deeper, and answer more accurately.
It is reported that Kimi K3.1 Pro shows particularly significant improvements in coding, 3D creation, financial research, and other long-horizon Agent capabilities. In multiple authoritative international evaluations, the comprehensive intelligence level of Kimi K3.1 Pro is already close to the world's leading closed-source models.
In the programming field, Kimi K3.1 Pro can skillfully use professional design and development tools and combine web search with 3D generation to create a complete experience, bringing more professional visual presentation to games and web pages. The pages and interactive scenes it produces show marked improvements in composition, layout, animation effects, and completeness, and its long-horizon coding capability in particular has been further enhanced through large-scale reinforcement learning.
In 3D creation, given a text description, a sketch, or a reference image, the new model can infer the shape, proportions, and spatial relationships of an object, directly operate professional modeling software to create a usable 3D model, and further build it into a complete interactive scene.
In the financial field, from retrieving materials and cross-checking data across documents to building financial models and conducting profitability analysis, the model can deliver complete end-to-end results, surpassing the world's frontier closed-source models in two international evaluations.
Since the beginning of this year, Chinese AI large models have delivered outstanding performance globally. According to statistics from the global AI model aggregation platform OpenRouter, last week (October 12 to 18), the total global invocation volume of AI large models was 164 trillion tokens, and the weekly invocation volume of Chinese large models has exceeded that of the United States for 25 consecutive weeks, firmly ranking first in the world. At the same time, the top five in global invocation volume last week were all Chinese AI large models. (End)