Exclusive: Paying for frontier AI models buys 4-month head start at 5x the cost
Summary
A Mozilla report shows that the performance gap between expensive closed frontier AI models from US companies and cheaper open AI models from others has shrunk to about 4.4 months. Many organizations now use open models for most routine tasks because they are nearly as effective but much less costly, reserving closed models for complex jobs that need more advanced AI.Key Facts
- The performance gap between top US closed AI models and Chinese open models is now about 4.4 months in task complexity.
- Open models like Moonshot AI’s Kimi K3 cost about 30% of leading closed models but perform almost as well.
- Closed models are preferred for expert-level, high-intensity, or long-context tasks that take 8–12 hours.
- Open AI models can handle tasks up to about 7 hours long with similar reliability.
- Closed models come with extra features like compliance support, software tools, and accountability.
- Many companies, including DoorDash, use open models for everyday work and closed models for harder tasks.
- The improvement in open AI models is increasing their use, narrowing the price-performance gap.
- Open models allow downloads and running on personal computers but usually do not reveal full training details.
Read the Full Article
This is a fact-based summary from The Actual News. Click below to read the complete story directly from the original source.
