Mozilla: Performance Gap Between US and Chinese AI Models Shrinks to 4.4 Months

According to Mozilla's latest 'State of Open Source AI' report, the performance gap between top-tier closed-source US models and Chinese open-source weight models has narrowed to approximately 4.4 months. Models like Moonshot AI's Kimi K3 are closing the gap with Anthropic's Fable 5, offering similar performance at roughly 30% of the cost. While current closed-source models can handle 12-hour expert tasks compared to 7 hours for open-source alternatives, projections suggest open-source models will reach current closed-source capabilities within four months. Mozilla suggests that organizations should prioritize open-source models, reserving premium closed-source services only for specialized, high-intensity use cases.