Kimi K3.1 Pro will be leading open-weights model once its weights are published (artificialanalysis.ai)

The recently launched Kimi K3.1 Pro has made a significant impact on the Artificial Analysis Intelligence Index, achieving a score of 55 and positioning itself as the leading open-weights AI model once its weights are published. This model demonstrates remarkable improvements in agentic performance, scoring 1,846 Elo on GDPval-AA—an increase of 312 Elo from its predecessor, Kimi K3—trailing only behind Claude Opus 5.5 and Muse Wonder Full. K3.1 Pro also leads in Terminal-Bench 4.0 with a score of 67%, bolstering its reputation as a robust tool in various evaluations. Its significance lies not only in its exceptional scores but also in the narrowing competitive gap between open-weights and proprietary models. Currently, the gap on the Intelligence Index is set to decrease from 12 points to just 3 after the weights release, highlighting the growing capabilities of open-source models in the field of AI/ML. With a large context window of 1 million tokens and a competitive pricing structure, Kimi K3.1 Pro positions itself as a formidable option for developers and researchers in the AI landscape, promoting advancements in accessibility and performance efficiency.