Artificial Analysis Releases Open Model Openness Index

Artificial Analysis introduced the Openness Index to evaluate AI models' contribution to the ecosystem. This index quantifies a model's openness based on criteria such as whether model weights are released, dataset information, and the accessibility of training code and related tools.

The key evaluation items are as follows:

• Model Weights: Release status and usage rights of model weights
• Training Data: Information about and accessibility of training datasets
• Code & Tools: Availability of code and tools for running and fine-tuning the model

This index provides developers with an objective standard for choosing models based not just on performance (Benchmark) alone, but also on Reproducibility and customizability.