The A.I. Daily
Open Chinese models narrow gap with frontier models
A Mozilla report finds that the performance gap between Chinese open-weight models and US frontier models has narrowed to roughly 4.4 months. It highlights Moonshot AI's Kimi K3 scoring three points behind Anthropic's Fable 5 on a composite index while costing about 30% as much, suggesting open models are preferable for most routine workloads.