Exclusive: Paying for frontier AI models buys 4-month head start at 5x the cost
Ars previewed Mozilla’s report on how cheap open models caught up on capability.
The performance gap between frontier AI models developed by US tech companies and the leading open-source models from China has narrowed to just 4.4 months, according to a Mozilla report. This development has led many organizations to shift towards cheaper open-source models for their routine tasks, highlighting certain workloads where frontier models remain a worthwhile investment.
Typically, businesses should default to open models for the majority of their operations, suggests the latest State of Open Source AI report from Mozilla, which was released on September 15 and shared with Ars prior to publication. The report underscores how top open-source model, Moonshot AI’s Kimi K3, matches the performance of Anthropic’s Fable 5 closed frontier model in most tasks, yet costs only 30 percent of the latter.
According to Raffi Krikorian, Mozilla's chief technology officer, frontier models offer a premium only in expert professional work, high-intensity retrieval, and handling long context. The decision to pay for closed models should be made based on the specific workload requirements rather than being a blanket organization-wide choice.
Written by urgent.news from Ars Technica's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
