Kimi K3.1 Pro puts open-source AI a month behind the best (www.modelcompare.ai)

On October 23, Kimi launched its K3.1 Pro model, scoring 55.1 on the Artificial Analysis Intelligence Index—a notable advancement in the open-source AI landscape. While this score falls 2.5 points behind the leading Claude Opus 5.5 model, it surpasses Xiaomi's MiMo-V2.6-Pro by 8.8 points, marking a competitive step forward for Kimi. The weights for K3.1 Pro are set to be freely available by November 4, enabling developers to implement the model through Kimi's API in the interim. This release highlights a significant trend: Chinese labs have consistently dominated the open-source AI space, holding the top position for over two years. In contrast, the best open model available from outside China, Meta's Muse Spark Open, currently lags significantly with a score of 38.9. The introduction of Kimi K3.1 Pro brings international competition closer, reducing the gap to about three months. As Kimi continues to release competitive models, the open-source ecosystem is expected to thrive, providing more accessible and cost-effective AI options for developers worldwide.