This chart provides an overview of current developments in the field of AI models, focusing on closed and open-source models in terms of performance. The key questions addressed by the chart are:
Performance alignment between open-source and closed models
- Open-source models, marked in green, are getting closer and closer to closed models (red points) in performance, although there are still gaps, especially in the GPT-4 class.
- Continued rapid progress in AI or possible slowdown – If the current rate of progress continues, we can expect further improvements in both open and closed models. The question is whether current trends will be slowed by technological or research limitations.
- The Future of AI – This chart suggests that open-source models could reach or even surpass closed models in the future. This opens the door to wider availability of powerful AI tools, which could have a widespread impact across the industry.
The high performance in the GPT-4 class indicates that the latest models are delivering exceptional results, which means significant benefits for both business and research, but also highlights the potential imbalance between the availability of these models.
Src: Threads
April 2024