Kimi K3 Tops BAAI's OpenCompass as World's #1 Open-Source Model

The Beijing Academy of Artificial Intelligence (BAAI) has published updated OpenCompass leaderboard data showing that Moonshot AI's Kimi K3 has reached the #1 position globally among open-source models, surpassing previous leaders including DeepSeek V4 Pro and other heavyweight contenders. The ranking marks a new milestone for Chinese-developed open-source large language models on an internationally recognized evaluation platform.

Kimi K3 was officially unveiled on July 16, 2026, during the World Artificial Intelligence Conference (WAIC 2026). With 2.8 trillion total parameters, it is currently the largest open-source model by parameter count in the world. Built on a Mixture-of-Experts (MoE) architecture, it natively supports text-and-image multimodal input, offers a 1-million-token context window, and delivers strong performance across coding, knowledge reasoning, and comprehension tasks.

OpenCompass is a comprehensive open-source model evaluation platform developed by BAAI. It covers benchmarks across language understanding, mathematical reasoning, code generation, and knowledge Q&A, making it one of the most influential third-party evaluation frameworks in the open-source AI community. Kimi K3's #1 ranking on OpenCompass's open-source track reflects strong validation from an authoritative evaluator.

Moonshot AI has announced that Kimi K3's model weights will be released on July 27 under a Modified MIT license that permits commercial use. This open-source strategy positions Kimi K3 not just as a technical flagship but as infrastructure that enterprises and developers can deploy locally and build upon.

Since its launch, multiple third-party evaluators have given positive feedback. In real-world scenarios such as code generation, long-document comprehension, and logical reasoning, Kimi K3 has demonstrated capabilities approaching or matching those of cutting-edge closed-source models — a key inflection point for the open-source AI ecosystem.

From an industry perspective, Kimi K3's top ranking sends two clear signals. First, open-source models are rapidly closing the capability gap with proprietary frontier models, and in some dimensions have begun to overtake them. Second, Chinese AI teams are now competing head-to-head with global leaders in architectural innovation and engineering execution.

Key developments to watch include community adoption velocity after the July 27 weight release, whether competitors accelerate their own open-source releases to reclaim the top spot, and Kimi K3's commercial traction via API pricing and enterprise deployments. For developers and enterprises seeking locally deployable high-performance models, July 27 is a significant date on the calendar.