I'm not sure either. It seems that many companies are testing their advanced programs in laboratories with select researchers. Google, Microsoft, and pharmaceutical companies are all involved in this. Individual researchers are also creating their own agents using platforms that many AI models for a subscription fee.

My professor has a friend who uses Poe.com, which provides access to some of the best models at the cost of a ChatGPT Pro subscription. You can create agents that work very effectively and assign different models to them, such as Qwen, Claude, Liquide, Gemini, Palm, Deepseek, and more. I did tried their pro plan and created several teacher agents, which was a great experience. It's been a year, and I'm sure they have advanced even further. This might be one of the best ways to experience your agents.