The analysis estimates Chinese frontier models like Kimi-K3 are currently 4.37 to 5.29 months behind US frontier models, based on backward-looking data. Forward-looking projections suggest Chinese models will be 8.57 to 8.83 months behind Mythos-Preview by the end of December 2026, with the gap appearing to widen.
The quantified and widening capability gap between Chinese and US frontier models suggests a sustained lead for US labs, impacting global AI model competition.