Open-Weights Models Are Completely Erasing the Proprietary AI Lead

The latest Artificial Analysis Index shows open-weights models like GLM-5.2 hitting an intelligence score of 51; effectively matchingGPT-5.6 Luna and outperforming Gemini 3.5 Flash.

I think this rapid closing of the open-weights gap is the death knell for raw proprietary model API monopolies, as builders gain direct access to near-frontier reasoning without the cost or data lock-in.