Leading Chinese AI company Kimi is accelerating its global expansion. According to the latest news from 36 Kr, Kimi recently disclosed to investors that its overseas revenue has officially exceeded domestic revenue. Especially after the release of its new generation large model Kimi K2.5, the number of global paid users increased by an astonishing 4 times within a few days.
Strong Performance: K2.5 Joins the Global Top Tier of Models
In third-party large model evaluation platform OpenRouter, Kimi K2.5 performed remarkably, climbing to the third position globally, behind Claude 4.5 Sonnet and Gemini 1.5 Flash. This achievement not only proves Kimi's strong technical foundation but also lays the groundwork for its explosive growth in the global market.
Yang Zhilin, founder of Moonshot, pointed out that relying solely on internet data for prediction has reached its limits. Kimi K2.5 has taken a different approach, exploring a "Agent Cluster" model.
Parallel Processing Capability: K2.5 can simultaneously coordinate up to 100 agents (Agents), working together to handle 1500 complex steps.
Efficiency Leap: In tasks such as large-scale information collection, this "teamwork" mode increases traditional AI efficiency by 3 to 10 times.
With the release of K2.5, Kimi's development path has become clearer: on the model level, it targets top global labs like Anthropic, continuously enhancing global influence through technological open-sourcing; on the application side, it aims to be a powerful productivity tool similar to Manus.
Currently, Kimi has fully open-sourced model weights and toolchains starting from the K2 version, aiming to allow global developers to deploy Kimi's underlying capabilities locally or in the cloud. This dual-driven strategy of "basic model ceiling + ultimate productivity applications" is helping Kimi accelerate its pace in the global AI race.
