Open models narrow frontier AI performance gap

A Mozilla report reveals that the performance gap between US and Chinese frontier AI models has narrowed to just 4.4 months. The report, published on September 15, highlights that open models are increasingly favored for routine tasks due to their lower cost. Moonshot AI’s Kimi K3, an open model, scores just three points behind Anthropic’s Fable 5, while costing 30 percent less. Raffi Krikorian, Mozilla’s chief technology officer, stated that closed models justify their higher cost in specific workloads like expert tasks and long context processing. The report suggests most organizations should use open models as the default for most operations, with closed models reserved for specialized applications. The findings underscore a growing trend toward open-source AI solutions in the industry. The State of Open Source AI report emphasizes that while closed models maintain an edge in certain high-intensity tasks, their premium is workload-specific rather than organization-wide. This shift reflects broader industry moves toward cost-effective open models, which are becoming viable alternatives for a wide range of applications. Mozilla’s analysis provides a clear framework for evaluating when to invest in closed models versus adopting open-source solutions. The report’s insights are expected to influence corporate AI strategies in the coming months.