> DeepSeek can surpass proprietary models by “profiting” from open research and open source, why couldn’t the proprietary models do the same?

DeepSeek did something legitimately innovative with their addition of Group Relative Policy Optimization. Other firms are certainly free to innovate as well.